Network Working Group                                     R. R. Stewart
INTERNET-DRAFT                                                   Q. Xie
                                                               Motorola
                                                           K. Morneault
                                                               C. Sharp
                                                                  Cisco
                                                     H. J. Schwarzbauer
                                                                Siemens
                                                              T. Taylor
                                                        Nortel Networks
                                                              I. Rytina
                                                               Ericsson
                                                               M. Kalla
                                                              Telcordia
                                                               L. Zhang
                                                                   UCLA
                                                              V. Paxson
                                                                  ACIRI

expires in six months                                   October 15,1999 20,1999

                 Simple Control Transmission Protocol
                   <draft-ietf-sigtran-sctp-01.txt>
                   <draft-ietf-sigtran-sctp-02.txt>

Status of This Memo

This document is an Internet-Draft and is in full conformance with all
provisions of Section 10 of RFC2026. Internet-Drafts are working
documents of the Internet Engineering Task Force (IETF), its areas,
and its working groups. Note that other groups may also distribute
working documents as Internet-Drafts.

The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt

The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.

Abstract

This document describes the Simple Control Transmission Protocol
(SCTP). SCTP is designed to transport PSTN signalling messages over
IP networks, but is capable of broader application.

SCTP is an application-level datagram transfer protocol operating on
top of an unreliable datagram service such as UDP. It offers the
following services to its users:

  -- acknowledged error-free non-duplicated transfer of user data
  -- application-level segmentation to conform to discovered MTU size
  -- sequenced delivery of user datagrams within multiple streams,
     with an option for order-of-arrival delivery of individual
     datagrams
  -- optional multiplexing of user datagrams into SCTP datagrams,
     subject to MTU size restrictions
  -- enhanced reliability through support of multi-homing at either or
     both ends of the association.

The design of SCTP includes appropriate congestion avoidance behaviour
and resistance to flooding and masquerade attacks.

Stewart, et al                                               [Page  1]
                        TABLE OF CONTENTS

1.  Introduction
  1.1 Motivation
  1.2 Architectural View of SCTP
  1.3 Functional View of SCTP
    1.3.1 Association Startup and Takedown
    1.3.2 Sequenced Delivery within Streams
    1.3.3 User Data Segmentation
    1.3.4 Acknowledgement and Congestion Avoidance
    1.3.5 Chunk Multiplex
    1.3.6 Path Management
    1.3.7 Message Validation
  1.4 Recapitulation of Key Terms
  1.5. Abbreviations
2.  SCTP Datagram Format
  2.1 SCTP Common Header Field Descriptions
  2.2 Chunk Field Descriptions
    2.2.1  Optional/Variable-length Parameter Format
    2.2.2 Vendor-Specific Extension Parameter Format
  2.3 SCTP Chunk Definitions
    2.3.1 Initiation (INIT)
      2.3.1.1 Optional or Variable Length Parameters
    2.3.2 Initiation Acknowledgement (INIT ACK)
      2.3.2.1 Optional or Variable Length Parameters
    2.3.3 Selective Acknowledgement (SACK)
    2.3.4 Heartbeat Request (HEARTBEAT)
    2.3.5 Heartbeat Acknowledgment (HEARTBEAT ACK)
    2.3.6 Abort Association (ABORT)
    2.3.7 Shutdown Association (SHUTDOWN)
    2.3.8 Shutdown Acknowledgment (SHUTDOWN ACK)
    2.3.9 Operation Error (ERROR)
    2.3.10 Encryption Cookie (COOKIE)
    2.3.11 Cookie Acknowledgment (COOKIE ACK)
    2.3.12 Payload Data (DATA)
  2.4 Vendor-Specific Chunk Extensions
3. SCTP Association State Diagram
4. Association Initialization
  4.1 Normal Establishment of an Association
    4.1.1 Handle Stream Parameters
    4.1.2 Handle Address Parameters
    4.1.3 Generating Responder Cookie
    4.1.4 Cookie Processing
    4.1.5 Cookie Authentication
    4.1.6 An Example of Normal Association Establishment
  4.2 Handle Duplicate INIT, INIT ACK, COOKIE, and COOKIE ACK
    4.2.1 Handle Duplicate INIT in COOKIE-WAIT or COOKIE-SENT State
    4.2.2 Handle Duplicate INIT in Other States
    4.2.3 Handle Duplicate INIT ACK
    4.2.4 Handle Duplicate COOKIE
    4.2.5 Handle Duplicate COOKIE-ACK
    4.2.6 Handle Stale COOKIE Error
  4.3 Other Initialization Issues
    4.3.1 Selection of Tag Value
    4.3.2 Initiation from behind a NAT
5. User Data Transfer
  5.1 Transmission of DATA Chunks
  5.2 Acknowledgment of Reception of DATA Chunks
  5.3 Management Retransmission Timer
    5.3.1 RTO Calculation
    5.3.2 Retransmission Timer Rules
    5.3.3 Handle T3-rxt Expiration
  5.4 Multi-homed SCTP Endpoints
  5.5 Stream Identifier and Sequence Number
  5.6 Ordered and Un-ordered Delivery
  5.7 Report Gaps in Received DATA TSNs
  5.8 CRC-16 Utilization
  5.9 Segmentation
  5.10 Bundling and Multiplexing
6. Congestion Control
  6.1 SCTP Differences from TCP Congestion Control
  6.2 SCTP Slow-Start and Congestion Avoidance
    6.2.1 Slow-Start
    6.2.2 Congestion Avoidance
    6.2.3 Congestion Control
    6.2.4 Fast Retransmit on Gap Reports
  6.3 Path MTU Discovery
  6.4 Discussion
7.  Fault Management
  7.1 Endpoint Failure Detection
  7.2 Path Failure Detection
  7.3 Path Heartbeat
  7.4 Verification Tag
8. Termination of Association
  8.1 Close of an Association
  8.2 Shutdown of an Association
9. Interface with Upper Layer
  9.1 ULP-to-SCTP
  9.2 SCTP-to-ULP
  9.3 Interfaces to Layer Management
    9.3.1 LM-to-SCTP
    9.3.2 SCTP-to-LM
10. Security Considerations
  10.1 Security Objectives
  10.2 SCTP Responses To Potential Threats
    10.2.1 Countering Insider Attacks
    10.2.2 Protecting against Data Corruption in the Network
    10.2.3 Protecting Confidentiality
    10.2.4 Protecting against Blind Denial of Service Attacks
      10.2.4.1 Flooding
      10.2.4.2 Masquerade
      10.2.4.3 Improper Monopolization of Services
  10.3 Protection against Fraud and Repudiation
11. IANA Consideration
  11.1 IETF-defined Chunk Extension
  11.2 IETF-defined Chunk Parameter Extension
  11.3 IETF-defined Additional Error Causes
12. Suggested SCTP Timer and Protocol Parameter Values
13. Acknowledgments
14. Authors' Addresses
15. References

1. Introduction

This section explains the reasoning behind the development of the
Simple Control Transmission Protocol (SCTP), the services it offers,
and the basic concepts needed to understand the detailed description
of the protocol.

1.1 Motivation

TCP [RFC 793, "Transmission Control Protocol", Jon Postel ed.,
September 1981] has performed immense service as the primary means of
reliable data transfer in IP networks. However, an increasing number
of recent applications have found TCP too limiting, and have
incorporated their own reliable data transfer protocol on top of UDP
[RFC 768, "User Datagram Protocol", Jon Postel, August 1980]. The
limitations which users have wished to bypass relate both to the
intrinsic nature of TCP and to its typical implementation.

Intrinsic limitations:

     -- TCP provides both reliable data transfer and strict order-
     of-arrival delivery of data. Some applications need reliable
     transfer without sequence maintenance, while others would be
     satisfied with partial ordering of the data. In both of these
     cases the head-of-line blocking offered by TCP causes
     unnecessary delay.

     -- The stream-oriented nature of TCP is often an inconvenience.
     Applications must add their own record marking to delineate
     their messages, and must make explicit use of the push facility
     to ensure that a complete message is transferred in a
     reasonable time.

     -- The limited scope of TCP sockets complicates the task of
     providing highly-available data transfer capability using
     multi-homed hosts.

Limitations due to implementation:

     -- TCP is generally implemented at the operating system level.
     Kernel limitations may constrain the maximum allowable number
     of simultaneous TCP connections to a number far below that
     required for certain applications.

     -- TCP implementations do not generally allow the application
     to control the timers which determine how quickly a connection
     failure is discovered. Some applications are more critically
     dependent than others on timely initiation of recovery from
     such failures.

Transport of PSTN signalling across the IP network is an application
for which all of these limitations of TCP are relevant. While this
application directly motivated the development of SCTP, other
applications may find SCTP a good match to their requirements.

1.2 Architectural View of SCTP

SCTP is viewed as a layer between the SCTP user application ("SCTP
user" for short) and an unreliable end-to-end datagram service such as
UDP. The basic service offered by SCTP is the reliable transfer of
user datagrams between peer SCTP users. It performs this service
within the context of an association between two SCTP Hosts. Chapter 9
of this document sketches the API which should exist at the boundary
between the SCTP and the SCTP user layers.

SCTP is connection-oriented in nature, but the SCTP association is a
broader concept than the TCP connection.  SCTP provides the means for
each SCTP endpoint to provide the other during association startup
with a list of transport addresses (e.g. address/UDP port
combinations) through which that endpoint can be reached and from
which it will originate messages.  The association spans transfers
over all of the possible source/destination combinations which may be
generated from the two endpoint lists.

   _____________                                      _____________
  |  SCTP User  |                                    |  SCTP User  |
  | Application |                                    | Application |
  |-------------|                                    |-------------|
  |    SCTP     |                                    |    SCTP     |
  |  Transport  |                                    |  Transport  |
  |   Service   |                                    |   Service   |
  |-------------|                                    |-------------|
  |  Unreliable |One or more    ----      One or more| Unreliable  |
  |   Datagram  |port/address    \/      port/address|  Datagram   |
  |   Service   |appearances     /\       appearances|   Service   |
  |_____________|               ----                 |_____________|

    SCTP Host A |<-------- Network transport ------->| SCTP Host B

                    Figure 1: An SCTP Association

It is possible to have multiple associations active between the same
pair of SCTP Hosts.

1.3 Functional View of SCTP

The SCTP transport service can be decomposed into a number of
functions. These are depicted in Figure 2 and explained in the
remainder of this section.

                   SCTP User Application

  ..-----------------------------------------------------
  .. _____________                  ____________________
    |             |                | Sequenced delivery |
    | Association |                |   within streams   |
    |             |                |____________________|
    |   startup   |
  ..|             |         ____________________________
    |     and     |        |    User Data Segmentation  |
    |             |        |____________________________|
    |   takedown  |
  ..|             |         ____________________________
    |             |        |     Acknowledgement        |
    |             |        |          and               |
    |             |        |    Congestion Avoidance    |
  ..|             |        |____________________________|
    |             |
    |             |         ____________________________
    |             |        |       Chunk Multiplex      |
    |             |        |____________________________|
    |             |
    |             |     ________________________________
    |             |    |         Path Management        |
    |             |    |________________________________|
    |             |
    |             |     ________________________________
    |             |    |     Message Validation         |
    |______________    |________________________________|

   Figure 2: Functional View of the SCTP Transport Service

1.3.1 Association Startup and Takedown

An association is initiated by a request from the SCTP user (see the
description of the ASSOCIATE primitive in Chapter 9). The startup
sequence is described in chapter 4 of this document. It is designed to
be resistant to flooding and masquerade attacks.

SCTP provides for graceful takedown of an active association on
request from the SCTP user. See the description of the TERMINATE
primitive in chapter 9. SCTP also allows ungraceful takedown, either
on request from the user (ABORT primitive) or as a result of an error
condition detected within the SCTP layer. Chapter 8 describes both the
graceful and the ungraceful takedown procedures.

1.3.2 Sequenced Delivery within Streams

The term "stream" is used in SCTP to refer to a sequence of
datagrams. This is in contrast to its usage in TCP, where it refers to
a sequence of bytes.

The SCTP user can specify at association startup time the number of
streams to be supported by the association. This number is negotiated
with the remote end (see section 4.1.1). User datagrams are associated
with stream numbers (SEND, RECEIVE primitives, Chapter 9). Internally,
SCTP assigns a stream sequence number to each datagram passed to it by
the SCTP user. On the receiving side, SCTP ensures that datagrams are
delivered to the SCTP user in sequence within a given stream. However,
while one stream may be blocked waiting for the next in-sequence user
datagram, delivery from other streams may proceed.

SCTP provides a mechanism for bypassing the sequenced delivery
service. Usr datagrams sent using this mechanism are delivered to the
SCTP user as soon as they are received.

1.3.3 User Data Segmentation

SCTP can segment user datagrams to ensure that the SCTP datagram
passed to the lower layer conforms to the path MTU. Segments are
reassembled into complete datagrams before being passed to the SCTP
user.

1.3.4 Acknowledgement and Congestion Avoidance

SCTP assigns a Transmission Sequence Number (TSN) to each user data
segment or unsegmented datagram. The TSN is independent of any
sequence number assigned at the stream level. The receiving end
acknowledges all TSNs received, even if there are gaps in the
sequence. In this way, reliable delivery is kept functionally separate
from sequenced delivery.

The Acknowledgement and Congestion Avoidance function is responsible
for message retransmission when timely acknowledgement has not been
received. Message retransmission is conditioned by congestion
avoidance procedures similar to those used for TCP.

See Chapters 5 and 6 for a detailed description of the protocol
procedures associated with this function.

1.3.5 Chunk Multiplex

As described in Chapter 2, the SCTP datagram as delivered to the lower
layer consists of a common header followed by one or more chunks. Each
chunk may contain either user data or SCTP control information. The
SCTP user has the option to request "bundling", or multiplexing of
more than one user datagram into a single SCTP datagram. The chunk
multiplex function of SCTP is responsible for assembly of the complete
SCTP datagram and its disassembly at the receiving end.

1.3.6 Path Management

The sending SCTP user is able to manipulate the set of transport
addresses used as destinations for SCTP datagrams, through the
primitives described in Chapter 9. The SCTP path management function
chooses the destination transport address for each outgoing SCTP
datagram based on the SCTP user's instructions and the currently
perceived reachability status of the eligible destination set.

The path management function monitors reachability through heartbeat
messages where other message traffic is inadequate to provide this
information, and advises the SCTP user when reachability of any far-
end transport address changes. The path management function is also
responsible for reporting the eligible set of local transport
addresses to the far end during association startup, and for reporting
the transport addresses returned from the far end to the SCTP user.

At association start-up, a primary destination transport address is
defined for each SCTP endpoint, and is used for normal sending of SCTP
datagrams.

On the receiving end, the path management is responsible for verifying
the existence of a valid SCTP association to which the inbound SCTP
datagram belongs before passing it for further processing.

1.3.7 Message Validation

The common SCTP header includes a validation tag and an optional CRC
field. A validation tag value is chosen by each end of the association
during association startup. Messages received without the validation
tag value expected by the receiver are discarded, as a protection
against blind masquerade attacks and against stale datagrams from a
previous association.

The CRC may optionally be set by the sender, to provide additional
protection against data corruption in the network beyond that provided
by lower layers (e.g. the UDP checksum).

1.4 Recapitulation of Key Terms

The language used to describe SCTP has been introduced in the previous
sections. This section provides a consolidated list of the key terms
and their definitions.

 o SCTP user application (SCTP user): The logical higher-layer
   application entity which uses the services of SCTP, also called
   the Upper-layer Protocol (ULP).

 o User datagram (user message): the unit of data delivery across the
   interface between SCTP and its user.

 o User data: the content of user datagrams.

 o SCTP datagram: the unit of data delivery across the interface
   between SCTP and the unreliable datagram service (e.g. UDP) which
   it is using. An SCTP datagram includes the common SCTP header,
   possible SCTP control chunks, and user data encapsulated within
   SCTP DATA chunks.

 o SCTP host: a logical unit within which SCTP is running.

 o Transport address: an address which serves as a source or
   destination for the unreliable datagram transport service used by
   SCTP.  In IP networks, a transport address is defined by the
   combination of an IP address and a UDP port number.

 o SCTP endpoint: a logical entity, comprising a set of eligible
   transport addresses at an SCTP host, which SCTP datagrams will be
   sent to and received from. Note, a transport address can only be
   included in one unique SCTP endpoint, i.e., it is NOT allowed to
   have the same SCTP transport address appear in more than one
   endpoints.

 o SCTP association: a protocol relationship between SCTP endpoints,
   comprising the two SCTP endpoints and protocol state information
   including verification tags and the currently active set of
   Transmission Sequence Numbers (TSNs), etc.

 o Chunk: a unit of information within an SCTP datagram, consisting of
   a chunk header and chunk-specific content.

 o Transmission Sequence Number (TSN): a 32-bit sequence number used
   internally by SCTP. One TSN is attached to each chunk containing
   user data to permit the receiving SCTP endpoint to acknowledge its
   receipt and detect duplicate deliveries.

 o Stream: a uni-directional logical channel established from one to
   another associated SCTP endpoints, within which all user datagrams
   are delivered in sequence except for those submitted to the
   unordered delivery service.

   Note: The relationship between stream numbers in opposite
   directions is strictly a matter of how the applications use
   them. It is the responsiblity of the SCTP user to create these
   correlations if they are so desired.

 o Stream sequence number: a 16-bit sequence number used internally by
   SCTP to assure sequenced delivery of the user datagrams within a
   given stream.  One stream sequence number is attached to each user
   datagram.

 o Bundling: an optional multiplexing operation, whereby more than one
   user datagram may be carried in the same SCTP datagram.  Each user
   datagram occupies its own DATA chunk.

 o Outstanding TSN (at an SCTP endpoint): a TSN (and the associated DATA
   chunk) which have been sent by the endpoint but for which it has not
   yet received an acknowledgement.

 o Unacknowledged TSN (at an SCTP endpoint): a TSN (and the associated DATA
   chunk) which have been received by the endpoint but for which an
   acknowledgement has not yet been sent.

 o Receiver Window (rwnd): The most recently advertised receiver
   window, in number of octets. This gives an indication of the space
   available in the receiver's inbound buffer.

 o Congestion Window (cwnd): An SCTP variable that limits the data, in
   number of octets, a sender can send into the network before
   receiving an acknowledgment on a particular destination Transport
   address.

 o Slow Start Threshold (ssthresh): An SCTP variable. This is the
   threshold which the endpoint will use to determine whether to
   perform slow start or congestion avoidance on a particular destination
   transport address. Ssthresh is in number of octets.

 o Transmission Control Block (TCB): an internal data structure
   created by an SCTP endpoint for each of its existing SCTP
   associations to other SCTP endpoints. TCB contains all the status
   and operational information for the endpoint to maintain and manage
   the corresponding association.

1.5. Abbreviations

MD5    - MD5 Message-Digest Algorithm [4]

NAT    - Network Address Translation

RTO    - Retransmission Time-out

RTT    - Round-trip Time

RTTVAR - Round-trip Time Variation

SCTP   - Simple Control Transmission Protocol

SRTT   - Smoothed RTT

TCB    - Transmission Control Block

TLV    - Type-Length-Value Coding Format

TSN    - Transport Sequence Number

ULP    - Upper-layer Protocol

2.  SCTP Datagram Format

An SCTP datagram is composed of a common header and chunks. A chunk
contains either control information or user data.

The SCTP datagram format is shown below:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Common Header                             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                         Chunk #1                              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          ...                                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                         Chunk #n                              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Multiple chunks can be multiplexed into one UDP SCTP datagram up to
the MTU size except for the INIT, INIT ACK, and SHUTDOWN ACK
chunks. These chunks MUST not be multiplexed with any other chunk in a
datagram. See Section 5.10 for more details on chunk multiplexing.

If an user data message doesn't fit into one SCTP datagram it can be
segmented into multiple chunks using the procedure defined in Section
5.9.

2.1 SCTP Common Header Field Descriptions

                      SCTP Common Header Format

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   | Vers  |     Reserved        |C|    CRC-16                     |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                 Verification Tag                              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Version: 4 bits, u_int

  This field represents the version number of the SCTP protocol,
  and MUST be set to '0011'.

Verification Tag: 32 bit u_int

  The receiver of this datagram uses the Verification Tag to identify
  the association. On transmit, the value of this Verification Tag MUST
  be set to the value of the Initiate Tag received from the peer
  endpoint during the association initialization.

  For datagrams carrying the INIT chunk, the transmitter MUST set the
  Verification Tag to all 0's.  If the receiver receives a datagram
  with an all-zeros Verification Tag field, it checks the Chunk ID
  immediately following the common header.  If the Chunk Type is not
  INIT or SHUTDOWN ACK, the receiver MUST drop the datagram.

  For datagrams carrying the SHUTDOWN-ACK chunk, the transmitter
  SHOULD set the Verification Tag to the Initiate Tag received from
  the peer endpoint during the association initialization, if known.
  Otherwise the Verification Tag MUST be set to all 0's.

Reserved:

  Reserved bits MUST be set to 0 on transmit and should be ignored
  on reception.

C:  1 bit (Octet 2, Bit 8)

  When the C-bit is set to 1, the CRC-16 field contains the CRC-16
  (defined below).

  When the C-bit is set to 0, the CRC-16 field is not used and MUST be
  set to 0.

CRC-16: (Octets 3 & 4)

  When the C Bit is set to 1, this field MUST contain a CRC-16.
  The CRC-16 used is defined in Section 4.2 of ITU Recommendation
  Q.703 [2].

  Section 5.8 defines the use of CRC-16 in SCTP.

  IMPLEMENTATION NOTE: When the C bit is set to 0, an implementation
  MAY use the fixed value 0x30000000 as a sanity check on an inbound
  datagram. If the first long integer is not the fixed value the
  datagram MAY be discarded with no further processing.

2.2  Chunk Field Descriptions

The figure below illustrates the field format for the chunks to be
transmitted in the SCTP datagram. Each chunk is formatted with a Chunk
ID field, a Chunk-specific flag field, a Length field and a value
field.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Chunk ID    |Chunk  Flags   |      Chunk Length             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   \                                                               \
   /                    Value                                      /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk ID: 8 bits, u_int

  This field identifies the type of information contained in the chunk
  value field. It takes a value from 0x00 to 0xFF. The value of 0xFE
  is reserved for vendor-specific extensions. The value of 0xFF is
  reserved for future use as an extension field. Procedures for
  extending this field by vendors are defined in Section 2.4.

  The values of Chunk ID are defined as follows:

  ID Value    Chunk Type
  -----       ----------
  00000000  - Payload Data (DATA)
  00000001  - Initiation (INIT)
  00000010  - Initiation Acknowledgment (INIT ACK)
  00000011  - Selective Acknowledgment (SACK)
  00000100  - Heartbeat Request (HEARTBEAT)
  00000101  - Heartbeat Acknowledgment (HEARTBEAT ACK)
  00000110  - Abort (ABORT)
  00000111  - Shutdown (SHUTDOWN)
  00001000  - Shutdown Acknowledgment (SHUTDOWN ACK)
  00001001  - Operation Error (ERROR)
  00001010  - Responder Cookie (COOKIE)
  00001011  - Cookie Acknowledgement (COOKIE ACK)
  00001100 to 11111101 - reserved for future IETF usage
  11111110  - Vendor-specific chunk extensions
  11111111  - IETF-defined Chunk Extension

Chunk Flags: 8 bits

  The usage of these bits depends on the chunk type as given by the
  Chunk ID. Unless otherwise specified, they are set to zero on
  transmit and are ignored on receipt.

Chunk Length: 16 bits (u_int)

  This value represents the size of the chunk in octets including the
  Chunk ID,  Flags, Length and Value fields.  Therefore, if the Value
  field is zero-length, the Length field will be set to 0x0004.  The
  Length field does not include any padding.

Chunk Value:  variable length

  The Chunk Value field contains the actual information to be
  transferred in the chunk. The usage and format of this field
  is dependent on the Chunk ID. The Chunk Value field MUST be
  aligned on 32-bit boundaries. If the length of the chunk does not
  align on 32-bit boundaries, it is padded at the end with all zero
  octets.

SCTP defined chunks are described in detail in Section 2.3. The
guidelines for Vendor-Specific chunk extensions are discussed in
Section 2.4. And the guidelines for IETF-defined chunk extensions
can be found in Section 11.1 of this document.

2.2.1  Optional/Variable-length Parameter Format

The optional and variable-length parameters contained in a chunk
are defined in a Type-Length-Value format as shown below.

   0                   1                   2                   3
   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  |          Parameter Type       |       Parameter Length        |
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
  \                                                               \
  /                       Parameter Value                         /
  \                                                               \
  +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Parameter Type:  16 bit u_int

  The Type field is a 16 bit identifier of the type of parameter. It
  takes a value of 0x0000 to 0xFFFF.

  The value of 0xFFFE is reserved for vendor-specific extensions if
  the specific chunk allows such extensions. The value of 0xFFFF is
  reserved for IETF-defined extensions.  Values other than those
  defined in specific SCTP chunk description are reserved for use by
  IETF.

Parameter Length:  16 bit u_int

  The Length field contains the size of the parameter in octets,
  including the Type, Length, and Value fields.  Thus, a parameter
  with a zero-length Value field would have a Length field of
  0x0004. The Length does not include any padding octets.

Parameter Value: variable-length.

  The Value is dependent on the value of the Type field.  The value
  field MUST be aligned on 32-bit boundaries.  If the value field is
  not aligned on 32-bit boundaries it is padded at the end with all
  zero octets.  The value field must be an integer number of octets.

The actual SCTP parameters are defined in the specific SCTP chunk
section. The guidelines for vendor-specific parameter extensions are
discussed in Section 2.2.2. And the rules for IETF-defined parameter
extensions are defined in Section 11.2.

2.2.2 Vendor-Specific Extension Parameter Format

This is to allow vendors to support their own extended parameters not
defined by the IETF. It MUST not affect the operation of SCTP.

Endpoints not equipped to interpret the vendor-specific information
sent by a remote endpoint MUST ignore it (although it may be
reported). Endpoints that do not receive desired vendor-specific
information SHOULD make an attempt to operate without it, although
they may do so (and report they are doing so) in a degraded mode.

A summary of the Vendor-Specific Extension format is shown below. The
fields are transmitted from left to right.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |    Parameter Type = 0xFFFE    |      Parameter Length         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Vendor-Id                                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   \                                                               \
   /                    Value                                      /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   Type: 16 bit u_int

      0xFFFE for all Vendor-Specific parameters.

   Length: 16 bit u_int

      Indicate the size of the parameter in octets, including the
      Type, Length, Vendor-Id, and Value fields.

   Vendor-Id: 32 bit u_int

      The high-order octet is 0 and the low-order 3 octets are the
      SMI Network Management Private Enterprise Code of the Vendor
      in network byte order, as defined in the Assigned Numbers (RFC
      1700).

   Value: variable length

      The Value field is one or more octets.  The actual format of the
      information is site or application specific, and a robust
      implementation SHOULD support the field as undistinguished
      octets.

      The codification of the range of allowed usage of this field is
      outside the scope of this specification.

      It SHOULD be encoded as a sequence of vendor type / vendor length
      / value fields, as follows.  The parameter field is
      dependent on the vendor's definition of that attribute.  An
      example encoding of the Vendor-Specific attribute using this
      method follows:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |    Parameter Type = 0xFFFE    |      Parameter Length         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Vendor-Id                                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |          VS-Type              |             VS-Length         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   /                    VS-Value                                   /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   VS-Type: 16 bit u_int

     This field identifies the parameter included in the VS-Value field.
     It is assigned by the vendor.

   VS-Length: 16 bit u_int

     This field is the length of the vendor-specific parameter and
     Includes the VS-Type, VS-Length and VS-Value (if included) fields.

   VS-Value:  Variable Length

     This field contains the parameter identified by the VS-Type field.
     It's meaning is identified by the vendor.

2.3 SCTP Chunk Definitions

This section defines the format of the different chunk types.

Note: integers in the chunk MUST be transmitted in network octet-order.

2.3.1 Initiation (INIT) (00000001)

This chunk is used to initiate a SCTP association between
two endpoints. The format of the INIT message is shown below:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |0 0 0 0 0 0 0 1|Chunk  Flags   |      Chunk Length             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                         Initiate Tag                          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Receive Window Credit                      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Number of Outbound Streams   |        (Reserved)             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Number of Inbound Streams    |        (Reserved)             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                          Initial TSN                          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   \                                                               \
   /              Optional/Variable-Length Parameters              /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

The INIT chunk contains the following parameters. Unless otherwise
noted, each parameter MUST only be included once in the INIT chunk.

Fixed Parameters                     Status
----------------------------------------------
Initiate Tag                        Mandatory
Receiver Window Credit              Mandatory
Number of Outbound Streams          Mandatory
Number of Inbound Streams           Mandatory
Initial TSN                         Mandatory

Variable Parameters                  Status     Type Value
-------------------------------------------------------------
IPv4 Address/Port (Note 1)          Optional    0x0005
IPv6 Address/Port (Note 1)          Optional    0x0006
Cookie Preservative                 Optional    0x0009

Note 1: The INIT chunks may contain multiple addresses that may be
IPv4 and/or IPv6 in any combination.

The sequence of parameters within an INIT may be processed in any
order.

Vendor-specific parameters are allowed in INIT. However, they MUST be
appended to the end of the above INIT chunk. The format of the
vendor-specific parameters MUST follow the Type-Length-value format as
defined in Section 2.2.2. In case an endpoint does not support the
vendor-specific data received, it MUST ignore the additional fields.

Initiate Tag: 32 bit u_int

  The receiver of the INIT (the responding end) records the value of
  the Initiate Tag parameter. This value MUST be placed into the
  Verification Tag field of every SCTP datagram that the responding end
  transmits within this association.

  The valid range for Initiate Tag is from 0x1 to 0xffffffff. See
  Section 4.3.1 for more on selection of the tag value.

  If the value of the Initiate Tag in a received INIT chunk is found
  to be 0x0, the receiver MUST treat it as an error and silently
  discard the datagram (or send Abort with tag=0??!).

Receive Window Credit (rwnd): 32 bit u_int

  This field defines the maximum number of octets of outbound data the
  receiver of the INIT is allowed to have outstanding (i.e. sent and
  not acknowledged).

Number of Outbound Streams (OS):  16 bit u_int

  Defines the number of outbound streams the sender of this INIT chunk
  wishes to create in this association. The value of 0 MUST NOT be
  used.

Number of Inbound Streams (MIS) : 16 bit u_int

  Defines the maximum number of streams the sender of this INIT chunk
  allows the peer end to create in this association. The value 0 MUST
  NOT be used.

Initial TSN (I-TSN) : 32 bit u_int

  Defines the initial TSN that the sender will use. This field MAY be
  set to the value of the Initiate Tag field.

The Reserved fields must be set to all 0 by the sender and ignored by
the receiver.

2.3.1.1  Optional or Variable Length Parameters

The following parameters follow the Type-Length-Value format as
defined in Section 2.2.1. The IP address fields MUST come after the
fixed-length fields.

Any extensions MUST come after the IP address fields.

IPv4 Address/Port

  This parameter contains an IPv4 address/port for use as a destination
  transport address by the receiver.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1|0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                   IPv4 Address                                |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Port                  |      Padding = 0              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  IPv4 Address: 32bit u_int

    Contains an IPv4 address of this endpoint. It is binary encoded.

  Port Number: 16 bit u_int

    Contains the UDP port number which the sender of this INIT wants
    to use for this address.

  Padding: 16 bits

    This field is set to 0x00 on transmit and ignored on receive.

IPv6 Address/Port:

    This parameter contains an IPv6 address/port for use as a
    destination transport address by the receiver.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0|0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               |
   |                   IPv6 Address                                |
   |                                                               |
   |                                                               |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Port                  |      Padding = 0              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  IPv6 Address: 128 bit u_int

    Contains an IPv6 address of the sender of this message. It is
    binary encoded.

  Port Number: 16 bit u_int

    Contains the UDP port number which the sender of this INIT wants
    to use for this address.

  Padding: 16 bits

    This field is set to 0x00 on transmit and ignored on receive.

  The values passed in the IPv4 and IPv6 Address/Port parameters
  indicate to the other end of the association which transport
  addresses this end will support for the association being
  initiated. Within the association, any one of these addresses may
  appear in the source address field of a datagram sent from this (the
  initiating) end, and may be used as a destination of a datagram sent
  from the other (the responding) end.

  Note that an endpoint MAY be multi-homed. A multi-homed endpoint may
  have access to different types of network, thus more than one
  address type may be present in one INIT chunk, i.e., IPv4 and IPv6
  addresses are allowed in the same INIT message.

  More than one IP Address parameter can be included in an INIT chunk.

  If the INIT contains a least one IP Address parameter, then only the
  transport addresses provided within the INIT may be used as
  destinations by the responding end. If the INIT does not contain any
  IP Address parameters, the responding end MUST use the source
  address associated with the received SCTP datagram as its sole
  destination address for the session.

Cookie Preservative

  The sender of the INIT shall use this parameter to suggest to the
  receiver of the INIT for a longer life-span of the Responder Cookie.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0|0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0|
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |         Suggested Cookie Life-span Increment (msec.)          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  Suggested Cookie Life-span Increment: 32bit u_int

  This parameter indicates to the receiver the need for a cookie that
  expires in a longer period of time than that of the previous one. It
  is normally added to the INIT message during the second attempt of
  establishing an association with a peer after the first attempt
  failed due to a Stale COOKIE report from the same peer. It is
  optional for the receiver to honor the suggested cookie life-span
  increment based upon its local security requirements.

2.3.2 Initiation Acknowledgement (INIT ACK) (00000010):

The INIT ACK chunk is used to acknowledge the initiation of a SCTP
association.

The parameter part of INIT ACK is formatted similarly to the INIT
chunk. It uses two extra variable parameters: The Responder Cookie
and the Unrecognized Parameter:

The format of the INIT ACK message is shown below:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |0 0 0 0 0 0 1 0|Chunk  Flags   |      Chunk Length             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Initiate Tag                              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                 Receive Window Credit                         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Number of Outbound Streams   |        (Reserved)             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |  Number of Inbound Streams    |        (Reserved)             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                 Initial TSN                                   |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   \                                                               \
   /        Optional/Variable-Length Parameters                    /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

The INIT ACK contains the following parameters. Unless otherwise
noted, each parameter MUST only be included once in the INIT ACK chunk.

Fixed Parameters                     Status
----------------------------------------------
Initiate Tag                        Mandatory
Receiver Window Credit              Mandatory
Number of Outbound Streams          Mandatory
Number of Inbound Streams           Mandatory
Initial TSN                      Mandatory

Variable Parameters                  Status     Type Value
-------------------------------------------------------------
Responder Cookie                    Mandatory   0x0007
IPv4 Address (Note 1)               Optional    0x0005
IPv6 Address (Note 1)               Optional    0x0006
Unrecognized Parameters             Optional    0x0008

Note 1: The INIT ACK chunks may contain multiple addresses that
may be IPv4 and/or IPv6 in any combination.

The Responder Cookie and Unrecognized Parameters use the Type-Length-
Value format as defined in Section 2.2.1 and are described below.  The
other fields are defined the same as their counterparts in the INIT
message.

2.3.2.1 Optional or Variable Length Parameters

Responder Cookie: variable size, depending on Size of Cookie

  This field MUST contain all the necessary state and parameter
  information required for the sender of this INIT ACK to create the
  association, along with an MD5 digital signature (128-bit). See
  Section 4.1.3 for details on Cookie definition. The Cookie MUST be
  padded with '0' to the next 32-bit word boundary; otherwise, the
  format of the Cookie is implementation-specific.

Unrecognized Parameters: Variable Size.

  This parameter is returned to the originator of the INIT message if
  the receiver does not recognize one or more Optional/Variable-length
  parameters in the INIT chunk. This parameter field will contain the
  unrecognized parameters copied from the INIT message complete
  with TLV.

Vendor-Specific parameters are allowed in INIT ACK. However, they
MUST be defined using the format described in Section 2.2.2, and be
appended to the end of the INIT ACK chunk. In case the receiver of
the INIT ACK does not support the vendor-specific parameters received,
it MUST ignore those fields.

2.3.3 Selective Acknowledgement (SACK) (00000011):

This chunk is sent to the remote endpoint to acknowledge received DATA
chunks and to inform the remote endpoint of gaps in the received
subsequences of DATA chunks as represented by their TSNs.

The SACK MUST contain the Highest Consecutive TSN ACK and Rcv Window
Credit (rwnd) parameters. By definition, the value of the Highest
Consecutive TSN ACK parameter is the last TSN received at the time the
Selective ACK is sent, before a break in the sequence of received TSNs
occurs; the next TSN value following this one has not yet been
received at the reporting end.  This parameter therefore acknowledges
receipt of all TSNs up to and including the value given.

The Selective ACK also contains zero or more fragment reports. Each
fragment report acknowledges a subsequence of TSNs received following
a break in the sequence of received TSNs.  By definition, all TSNs
acknowledged by fragment reports are higher than the value of the
Highest Consecutive TSN ACK.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 0 0 1 1|Chunk  Flags   |      Chunk Length             |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |                  Highest Consecutive TSN ACK                  |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |                   Rcv Window Credit (rwnd)                    |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |    Number of Fragments = N    |         (Reserved)            |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |     Fragment #1 Start         |   Fragment #1 End             |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    /                                                               /
    \                              ...                              \
    /                                                               /
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |     Fragment #N Start         |   Fragment #N End             |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags:

  Set to all zeros on transmit and ignored on receipt.

Highest Consecutive TSN ACK: 32 bit u_int

  This parameter contains the TSN of the last DATA chunk received in
  sequence before a gap. Therefore, it must be lower than the
  Start TSN of the first fragment, if present.

Receive Window Credit (rwnd): 32 bit u_int

  This field defines the new maximum number of octets of outbound data
  the receiver of this SACK is allowed to have outstanding (i.e. sent
  and not acknowledged).

Number of Fragments: 16 bit u_int

  Indicates the number of TSN fragments included in this Selective
  ACK.

Reserved: 16 bit

  Must be set to all 0 by the sender and ignored by the receiver.

Fragments:

  These fields contain the ack fragments. They are repeated for each
  fragment up to the number of fragments defined in the Number of
  Fragments field. All DATA chunks with TSNs between the (Highest
  Consecutive TSN + Fragment Start) and (Highest Consecutive TSN +
  Fragment End) of each fragment are assumed to have been received
  correctly.

Fragment Start: 16 bit u_int

  Indicates the Start offset TSN for this fragment. To calculate the
  actual TSN number the Highest Consecutive TSN is added to this
  offset number to yield the TSN. This calculated TSN identifies
  the first TSN in this fragment that has been received.

Fragment End:  16 bit u_int

  Indicates the End offset TSN for this fragment. To calculate the
  actual TSN number the Highest Consecutive TSN is added to this
  offset number to yield the TSN. This calculated TSN identifies
  the TSN of the last DATA chunk received in this fragment.

For example, assume the receiver has the following datagrams newly
arrived at the time when it decides to send a Selective ACK,

                 ----------
                 | TSN=17 |
                 ----------
                 |        | <- still missing
                 ----------
                 | TSN=15 |
                 ----------
                 | TSN=14 |
                 ----------
                 |        | <- still missing
                 ----------
                 | TSN=12 |
                 ----------
                 | TSN=11 |
                 ----------
                 | TSN=10 |
                 ----------

then, the parameter part of the Selective ACK MUST be constructed as
follows (assuming the new rwnd is set to 0x1234 by the sender):

        +---------------+--------------+
        | Highest Consecutive TSN = 12 |
        ----------------+---------------
        |         rwnd = 0x1234        |
        ----------------+---------------
        | num of frag=2 |   (rev = 0)  |
        ----------------+---------------
        |frag #1 strt=2 |frag #1 end=3 |
        ----------------+---------------
        |frag #2 strt=5 |frag #2 end=5 |
        --------------------------------

2.3.4 Heartbeat Request (HEARTBEAT) (00000100):

An endpoint should send this chunk to its peer endpoint of the current
association to probe the reachability of a particular destination
transport address defined in the present association.

The parameter fields MUST contain the time values.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 0 1 0 0|Chunk  Flags   |0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0|
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |                      Time Value 1 (e.g., sec)                 |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |                      Time Value 2 (e.g., usec)                |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags:

  Set to zero on transmit and ignored on receipt.

Time Value 1:  32 bit u_int
Time Value 2:  32 bit u_int

  The Time Values contain a time value meaningful only to the
  sender. For example, Time Value 1 may be in units of seconds and
  Time Value 2 in units of microseconds. Their value should be set to
  the current time at which this Heartbeat Request is sent.

  IMPLEMENTATION NOTE: For most systems the value is normally set by
  doing a system call to get the current time.

2.3.5 Heartbeat Acknowledgment (HEARTBEAT ACK) (00000101):

An endpoint should send this chunk to its peer endpoint as a response
to a Heartbeat Request (see Section 7.3).

The parameter field MUST contain the time values.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 0 1 0 1|Chunk  Flags   |0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0|
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |                     Time Value 1 (e.g., sec)                  |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |                     Time Value 2 (e.g., usec)                 |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags:

  Set to zero on transmit and ignored on receipt.

Time Value 1:  32 bit u_int
Time Value 2:  32 bit u_int

  The values of these two field SHALL be copied from the time values
  contained in the Heartbeat Request to which this Heartbeat
  acknowledgement is responding.

2.3.6 Abort Association (ABORT) (00000110):

The ABORT chunk is sent to the peer of an association to terminate the
association. The Abort chunk has no parameters.

If an endpoint receives an INIT or INIT ACK missing a mandatory
parameter, it MUST send an ABORT message to its peer. It SHOULD
include a Operational Error chunk with the Abort chunk to specify
the reason.

If an endpoint receives an ABORT with a format error or for an
association that doesn't exist, it drops the chunk and ignores it.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 0 1 1 0|Chunk  Flags   |0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0|
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags:

  Set to zero on transmit and ignored on receipt.

2.3.7 SHUTDOWN (00000111):

An endpoint in an association MUST use this chunk to initiate a
graceful termination of the association with its peer.  This chunk has
the following format.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 0 1 1 1|Chunk  Flags   |0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0|
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |                 Highest Consecutive TSN ACK                   |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags:

  Set to zero on transmit and ignored on receipt.

Highest Consecutive TSN ACK: 32 bit u_int

  This parameter contains the TSN of the last chunk received in
  sequence before any gaps.

2.3.8 Shutdown Acknowledgment (SHUTDOWN ACK) (00001000):

This chunk MUST be used to acknowledge the receipt of the SHUTDOWN
chunk at the completion of the shutdown process, see Section 8.2 for
details.

The SHUTDOWN ACK chunk has no parameters.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 1 0 0 0|Chunk  Flags   |0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0|
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags:

  Set to zero on transmit and ignored on receipt.

Note: if the endpoint that receives the SHUTDOWN message does not have
a TCB or tag for the sender of the SHUTDOWN, the receiver SHALL still
respond. In such cases, the receiver SHALL send back a stand-alone
SHUTDOWN ACK chunk in an SCTP datagram with the Verification Tag field
of the common header filled with all '0's.

2.3.9 Operation Error (ERROR) (00001001):

This chunk is sent to the other endpoint in the association to notify
certain error conditions. It contains one or more error causes. It has
the following parameters:

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 1 0 0 1| Chunk  Flags  |           Length              |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    \                                                               \
    /                    one or more Error Causes                   /
    \                                                               \
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags:

  Set to zero on transmit and ignored on receipt.

Length:

  Set to the size of the chunk in octets, including the chunk header
  and all the Error Cause fields present.

Error causes are defined as variable-length parameters using the
format described in 2.2.1, i.e.:

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |           Cause Code          |       Cause Length            |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    /                    Cause-specific Information                 /
    \                                                               \
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Cause Code: 16 bit u_int

  Defines the type of error conditions being reported.

Cause Length: 16 bit u_int

  Set to the size of the parameter in octets, including the Cause Code,
  Cause Length, and Cause-Specific Information fields

Cause-specific Information: variable length

  This field carries the details of the error condition.

Currently SCTP defines the following error causes:

  Cause of error
  ---------------
  Invalid Stream Identifier

   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |     Cause Code=1              |      Cause Length=8           |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                     Stream Identifier                         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  Cause of error
  ---------------
  Missing Mandatory Parameter

   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |     Cause Code=2              |      Cause Length=8+N*2       |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                   Number of missing params=N                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Missing Param ID #1         |   Missing Param ID #2         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |   Missing Param ID #N-1       |   Missing Param ID #N         |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  Each missing mandatory parameter ID should be specified in the
  message.

  Cause of error
  --------------
  Stale Cookie Error: indicating the receiving of a valid cookie
  which is however expired.

   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |     Cause Code=3              |       Cause Length=8          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                 Measure of Staleness (msec.)                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

  The sender of this error cause MAY choose to report how long
  past expiration the cookie is, by putting in the Measure of
  Staleness field the difference, in microseconds, between the current
  time and the time the cookie expired. If the sender does not wish to
  provide this information it should set Measure of staleness to 0.

Guidelines for IETF-defined Error Cause extensions are discussed in
Section 11.3 of this document.

2.3.10 Encryption Cookie (COOKIE) (00001010):

This chunk is used only during the initialization of an association.
It is sent by the initiator of an association to its peer to complete
the initialization process. This chunk  MUST precede any DATA chunk
sent within the association, but MAY be bundled with one or more DATA
chunks in the same datagram.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 1 0 1 0|Chunk  Flags   |         Length                |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |                     Cookie                                    |
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags: 8 bit

  Set to zero on transmit and ignored on receipt.

Length: 16 bit u_int

  Set to the size of the chunk in octets, including the 4 octets of
  the chunk header and the size of the Cookie.

Cookie: variable size

  This field must contain the exact cookie received in a previous INIT
  ACK.

2.3.11 Cookie Acknowledgment (COOKIE ACK) (00001011):

This chunk is used only during the initialization of an association.
It is used to acknowledge the receipt of a COOKIE chunk.  This chunk
MUST precede any DATA chunk sent within the association, but MAY be
bundled with one or more DATA chunks in the same SCTP datagram.

     0                   1                   2                   3
     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    |0 0 0 0 1 0 1 1|Chunk  Flags   |0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0|
    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Chunk Flags:

  Set to zero on transmit and ignored on receipt.

2.3.12 Payload Data (DATA) (00000000):

The following format MUST be used for the DATA chunk:

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |0 0 0 0 0 0 0 0| Reserved|U|B|E|    Length                     |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                              TSN                              |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |      Stream Identifier S      |    Sequence Number n          |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   \                                                               \
   /                 User Data (seq n of Stream S)                 /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

Note: the TSN, stream identifier, and sequence number MUST be
transmitted in network byte order.

Reserved: 5 bits
  should be set to all '0's and ignored by the receiver.

U bit: 1 bit
  The (U)nordered bit, if set, indicates that this is an unordered
  data chunk, and there is NO Sequence Number assigned to this DATA
  chunk. Therefore, the receiver MUST ignore the Sequence Number
  field.

  After reassembly (if necessary), unordered data chunks MUST be
  dispatched to the upper layer by the receiver without any attempt of
  re-ordering.

  Note, if an unordered user message is segmented, each segment MUST
  have its U bit set to 1.

B bit: 1 bit

  The (B)eginning segment bit, if set, indicates the first segment of
  an SCTP user message.

E bit:  1 bit
  The (E)nding segment bit, if set, indicates the last segment of an
  SCTP user message.

A non-segmented user message shall have both the B and E bits set
to 1. Setting both B and E bits to 0 indicates a middle segment of a
multi-segment SCTP user message, as summarized in the following table:

       B E                  Description
    ============================================================
    |  1 0 | First piece of a segmented SCTP user message.     |
    +----------------------------------------------------------+
    |  0 0 | Middle piece of a segmented user message          |
    +----------------------------------------------------------+
    |  0 1 | Last piece of a segmented SCTP user message.      |
    +----------------------------------------------------------+
    |  1 1 | Un-segmented Message                              |
    ============================================================

Length:  16 bits (16 bit u_int)

  This field indicates the length of the DATA chunk in octets.  It
  includes the Type field, the Reserved field, the U and B/E bits, the
  Length field, TSN, the Stream Identifier, the Stream Sequence
  Number, and the User Data fields. It does not include any padding.

TSN : 32 bits (32 bit u_int)

  This value represents the TSN for this DATA chunk.

Stream Identifier S: 16 bit u_int

  Identifies the stream to which the following user data belongs.

Sequence Number n: 16 bit u_int

  This value presents the sequence number of the following user
  data within the stream S. Valid range is 0x0 to 0xFFFF.

  Note, when a user message is segmented by SCTP for transport, the
  same sequence number MUST be carried in each of the segments of
  the message.

User Data: variable length

  This is the payload user data. The implementation MUST pad the end
  of the data to a 32 bit boundary with 0 octets. Any padding should
  NOT be included in the length field.

2.4 Vendor-Specific Chunk Extensions

This Chunk type is available to allow vendors to support their own
extended data formats not defined by the IETF. It MUST not affect the
operation of SCTP.

Endpoints not equipped to interpret the vendor-specific chunk sent by
a remote endpoint MUST ignore it. Endpoints that do not receive
desired vendor specific information SHOULD make an attempt to operate
without it, although they may do so (and report they are doing so) in
a degraded mode.

A summary of the Vendor-Specific Chunk format is shown below.  The
fields are transmitted from left to right.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |      Type     |    Flags      |             Length            |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                    Vendor-Id                                  |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   \                                                               \
   /                    Value                                      /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

   Type: 8 bit u_int

      0xFE for all Vendor-Specific chunks.

   Flags: 8 bit u_int

      Vendor specific flags.

   Length: 16 bit u_int

      Size of this Vendor-Specific chunks in octets, including the Type,
      Flags, Length, Vendor-Id, and Value fields.

   Vendor-Id: 32 bit u_int

      The high-order octet is 0 and the low-order 3 octets are the SMI
      Network Management Private Enterprise Code of the Vendor in
      network byte order, as defined in the Assigned Numbers (RFC 1700).

   Value: Variable length

      The Value field is one or more octets.  The actual format of the
      information is site or application specific, and a robust
      implementation SHOULD support the field as undistinguished
      octets.

      The codification of the range of allowed usage of this field is
      outside the scope of this specification.

3. SCTP Association State Diagram

During the lifetime of an SCTP association, the SCTP endpoints
progress from one state to another in response to various events. The
events that may potentially advance an endpoint's state include:

  o SCTP user primitive calls, e.g., [open], [shutdown], [abort],

  o reception of INIT, COOKIE, ABORT, SHUTDOWN, etc. control
    chunks, or

  o some timeout events.

The state diagram in the figures below illustrates state changes,
together with the causing events and resulting actions. Note that some
of the error conditions are not shown in the state diagram. Full
description of all special cases should be found in the text.

Note, chunk names are given in all capital letters, while parameter
names have the first letter capitalized, e.g., COOKIE chunk type vs.
Cookie parameter.

                    -----          -------- (frm any state)
                  /       \      /  rcv ABORT      [abort]
 rcv INIT        |         |    |   ----------  or ----------
 --------------- |         v    v   delete TCB     snd ABORT
 generate Cookie  \    +---------+                 delete TCB
 snd INIT.ACK       ---|  CLOSED |
                       +---------+
                        /      \      [open]
                       /        \     ---------------
                      |          |    create TCB
                      |          |    snd INIT
                      |          |    strt init timer
     rcv valid COOKIE |          v
 (1) ---------------- |      +------------+
     create TCB       |      | COOKIE_WAIT| (2)
     snd COOKIE.ACK   |      +------------+
                      |          |
                      |          |    rcv INIT.ACK
                      |          |    -----------------
                      |          |    snd COOKIE
                      |          |    stop init timer
                      |          |    strt cookie timer
                      |          v
                      |      +------------+
                      |      | COOKIE_SENT| (3)
                      |      +------------+
                      |          |
                      |          |    rcv COOKIE.ACK
                      |          |    -----------------
                      |          |    stop cookie timer
                      v          v
                    +---------------+
                    |  ESTABLISHED  |
                    +---------------+

              (from any state except CLOSED)
                            |
                            |
                        /--------+--------\
    [shutdown]         /                   \
    -----------------  |                   |
    check outstanding  |                   |
    data chunks        |                   |
                       v                   |
                  +---------+              |
              (4) |SHUTDOWN |              | rcv SHUTDOWN
                  |PENDING  |              | ----------------
                  +---------+              |         x
                       |                   |
  No more outstanding  |                   |
  -------------------  |                   |
  snd SHUTDOWN         |                   |
  strt shutdown timer  |                   |
                       v                   v
                  +---------+        +-----------+
              (5) |SHUTDOWN |        | SHUTDOWN  |  (5)
                  |SENT     |        | RECEIVED  |
                  +---------+        +-----------+
                       |                   |
  rcv SHUTDOWN.ACK     |                   |         x
  -------------------  |                   |-----------------
  stop shutdown timer  |                   | retransmit missing DATA
  delete TCB           |                   | send SHUTDOWN.ACK
                       |                   |    delete TCB
                       |                   |
                       \    +---------+    /
                        \-->| CLOSED  |<--/
                            +---------+

Note:

(1) If the received COOKIE is invalid (i.e., failed to pass the
    authentication check), the receiver MUST silently discard the
    datagram. Or, if the received COOKIE is expired (see Section
    4.1.5), the receiver SHALL send an ERROR chunk back. In
    either case, the receiver SHALL stay in the closed state.

(2) If the init timer expires, the endpoint SHALL retransmit INIT
    and re-start the init timer without changing state. This SHALL be
    repeated up to 'Max.Init.Retransmit' times. After that, the
    endpoint SHALL abort the initialization process and report the
    error to SCTP user.

(3) If the cookie timer expires, the endpoint SHALL retransmit
    COOKIE and re-start the cookie timer without changing
    state. This SHALL be repeated up to 'Max.Init.Retransmit'
    times. After that, the endpoint SHALL abort the initialization
    process and report the error to SCTP user.

(4) In SHUTDOWN-SENT state the endpoint SHALL acknowledge any received
    DATA chunks without delay

(5) In SHUTDOWN-RECEIVED state, the endpoint MUST NOT accept any new
    send request from its SCTP user.

4. Association Initialization

Before the first data transmission can take place from one SCTP
endpoint ("A") to another SCTP endpoint ("Z"), the two endpoints must
complete an initialization process in order to set up an SCTP
association between them.

The SCTP user at an endpoint SHOULD use the ASSOCIATE primitive to
initialize an SCTP association to another SCTP endpoint.

  IMPLEMENTATION NOTE: From an SCTP-user's point of view, an
  association may be implicitly opened, without an Associate primitive
  (see 9.1 B) being invoked, by the initiating endpoint's sending of
  the first user data to the destination endpoint. The initiating SCTP
  will assume default values for all mandatory and optional parameters
  for the INIT/INIT ACK.

Once the association is established, unidirectional streams will be
open for data transfer on both ends (see Section 4.1.1).

A cookie mechanism is employed during the initialization to provide
protection against security attacks. The cookie mechanism uses a
four-way handshaking, but the last two legs of which are allowed to
carry user data for fast setup.

4.1 Normal Establishment of an Association

The initialization process consists of the following steps (assuming
that SCTP endpoint "A" tries to set up an association with SCTP
endpoint "Z" and "Z" accepts the new association):

A) "A" shall first send an INIT message to "Z". In the INIT, "A" must
   provide its security tag "Tag_A" in the Initiate Tag field. Tag_A
   shall be a random number in the range of 0x1 to 0xffffffff (see
   4.3.1 for Tag value selection). After sending the INIT, "A" enters
   the COOKIE-WAIT state.

B) "Z" shall respond immediately with an INIT ACK message. In the
   message, besides filling in other parameters, "Z" must set the
   Verification Tag field to Tag_A, and also provide its own security
   tag "Tag_Z" in the Initiate Tag field.

   Moreover, "Z" shall generate and send along with the INIT ACK a
   responder cookie. See Section 4.1.3 for responder cookie
   generation.

   Note: after sending out INIT ACK with the cookie, "Z" should not
   allocate any resources, nor keep any states for the new
   association. Otherwise, "Z" will be vulnerable to resource attacks.

C) Upon reception of the INIT ACK from "Z", "A" shall leave COOKIE-WAIT
   state and enter the COOKIE-SENT state. "A" now sends the cookie
   received in the INIT ACK message in a cookie chunk. The cookie
   chunk can be bundled with any pending DATA chunks, but it MUST
   be the first chunk in the datagram.

D) Upon reception of the COOKIE chunk, Endpoint "Z" will reply with
   a COOKIE-ACK chunk after building a TCB and marking itself to
   the established state. A COOKIE-ACK chunk may also be combined with
   any pending DATA chunks (and/or SACK chunks), but the COOKIE-ACK
   chunk must be the first chunk in the datagram.

  IMPLEMENTATION NOTE: a implementation may choose to send the
  communication up primitive to the SCTP user upon reception
  of a valid cookie.

E) Upon reception of the COOKIE ACK, endpoint "A" will move from the
   COOKIE-SENT state to the ESTABLISHED state, stopping its INIT
   timer.

  IMPLEMENTATION NOTE: a implementation may choose to send the
  communication up primitive to the SCTP user upon reception
  of a the COOKIE ACK.

Note: no DATA chunk shall be carried in the INIT or INIT ACK message.

Note: if an endpoint receives an INIT, INIT ACK, or COOKIE chunk but
decides not to establish the new association due to lack of resources,
etc., it shall respond to the chunk with an ABORT chunk. The
Verification Tag field of the common header must be set to equal the
Initiate Tag value of the peer.

Note: After the reception of the first data chunk in an association
the receiver MUST immediately respond with a SACK to acknowledge
the data chunk, subsequent acknowledgements should be done as
described in section 5.2.

Note: When a SCTP endpoint sends a INIT or INIT ACK it MUST include
all of its transport addresses in the parameter section. This is
because it may NOT be possible to control the "sending" address that
a receiver of a SCTP datagram sees. A receiver thus MUST know every
address that may be a source address for a peer SCTP endpoint, this
assures that the inbound SCTP datagram can be matched to the proper
association.

4.1.1 Handle Stream Parameters

In the INIT and INIT ACK messages, the sender of the message shall
indicate the number of outbound streams (OS) it wishes to have in the
association, as well as the maximal inbound streams (MIS) it will
accept from the other endpoint.

After receiving these stream configuration information from the other
side, each endpoint shall perform the following check: if the
peer's MIS is less than the endpoint's OS, meaning that the peer is
incapable of supporting all the outbound streams the endpoint wants to
configure, the endpoint MUST either settle with MIS outbound streams,
or abort the association and report to its upper layer the resources
shortage at its peer.

After the association is initialized, the valid outbound stream
identifier range for either endpoint shall be 0 to
min(local OS, remote MIS)-1.

4.1.2 Handle Address Parameters

During the association initialization, an endpoint shall use the
following rules to discover and collect the destination transport
address(es) to its peer.

On reception of an INIT or INIT ACK message, the receiver shall record
any transport addresses specified as parameters in the INIT or INIT
ACK message, and use only these addresses as destination transport
addresses when sending subsequent datagrams to its peer. If NO
destination transport addresses are specified in the INIT or INIT ACK
message, then the source address from which the message arrived should
be used as the destination transport address for all datagrams.

4.1.3 Generating Responder Cookie

When sending an INIT ACK as a response to an INIT message, the sender
of INIT ACK should create a responder cookie and send it as part of
the INIT ACK. Inside this responder cookie, the sender should include
a security signature, a time stamp on when the cookie is created, and
the lifespan of the cookie, along with all the information necessary
for it to establish the association.

The following steps SHOULD be taken to generate the cookie:

1) create an association TCB using information from both the received
   INIT and the outgoing INIT ACK messages,

2) in the TCB, set the creation time to the current time of day, and
   the lifespan to the protocol parameter 'Valid.Cookie.Life',

3) attach a private security key to the TCB and generate a 128-bit MD5
   signature from the key/TCB combination (see [4] for details on
   MD5), and

4) generate the responder cookie by combining the TCB and the
   resultant MD5 signature.

After sending the INIT ACK with the cookie, the sender SHOULD delete
the TCB and any other local resource related to the new association,
so as to prevent resource attacks.

The private key should be a cryptographic quality random number with
a sufficient length. Discussion in RFC-1750 [1] can be helpful in
selection of the key.

4.1.4 Cookie Processing

When a cookie is received from its peer in an INIT ACK message, the
receiver of the INIT ACK MUST immediately send a COOKIE chunk to its
peer and MAY piggy-back any pending DATA chunks on the outbound
cookie chunk. The sender should also start a timer, and retransmit
the cookie chunk until a COOKIE ACK is received or the endpoint
is marked unreachable.

4.1.5 Cookie Authentication

When an endpoint receives a COOKIE chunk from another endpoint with
which it has no association, it shall take the following actions:

1) compute an MD5 signature using the TCB data carried in the cookie
   along with the receiver's private security key,

2) authenticate the cookie by comparing the computed MD5 signature
   against the one carried in the cookie. If this comparison fails,
   the datagram, including the COOKIE and the attached user data,
   should be silently discarded,

3) compare the creation time stamp in the cookie to the current local
   time, if the elapsed time is longer than the lifespan carried in
   the cookie, then the datagram, including the COOKIE and the
   attached user data, SHOULD be discarded and the endpoint MUST
   transmit a stale cookie operational error to the sending endpoint,

4) if the cookie is valid, create an association to the sender of the
   COOKIE message with the information in the TCB data carried in the
   COOKIE, and enter the ESTABLISHED state,

5) acknowledge any DATA chunk in the datagram following the rules
   defined in Section 5.2, and,

6) send a COOKIE ACK chunk to the sender acknowledging reception of
   the cookie. The COOKIE ACK MAY be piggybacked with any DATA chunk or
   SACK chunk (if a DATA chunk is present in the received datagram a
   SACK MUST be sent in the acknowledgement).

Note that if a COOKIE is received from an endpoint with which the
receiver of the COOKIE has an existing association, the proceedures in
section 4.2 should be followed.

4.1.6 An Example of Normal Association Establishment

In the following example, "A" initiates the association and then sends
a user datagram to "Z", then "Z" sends two user datagrams to "A"
later:

Endpoint A                                          Endpoint Z

{app sets association with Z}
INIT [INIT Tag=Tag_A
      & other info]  --------\
(Start T1-init timer)         \
(Enter COOKIE-WAIT state)      \---> (compose temp TCB and Cookie_Z)

                                /--- INIT ACK [Veri Tag=Tag_A,
                               /               INIT Tag=Tag_Z,
(Cancel T1-init timer, <------/                Cookie_Z, & other info]
                                     (destroy temp TCB)
COOKIE[ Cookie_Z] -----------\
(Start T1-init timer)         \
(Enter COOKIE-SENT state)      \---> (build TCB enter established state)

                               /---- COOKIE-ACK
                              /
(Cancel T1-init timer, <-----/
 Enter established state)
...
{app sends 1st user data; strm 0}
DATA [TSN=initial TSN_A
    Strm=0,Seq=1 & user data]--\
(Start T3-rxt timer)            \
                                 \->

                              /----- SACK [TSN ACK=init TSN_A,Frag=0]
(Cancel T3-rxt timer) <------/
...
                                     ...
                                     {app sends 2 datagrams;strm 0}
                               /---- DATA
                              /        [TSN=init TSN_Z
                          <--/          Strm=0,Seq=1 & user data 1]
SACK [TSN ACK=init TSN_Z,      /---- DATA
      Frag=0]      --------\  /        [TSN=init TSN_Z +1,
                            \/          Strm=0,Seq=2 & user data 2]
                     <------/\
                              \
                               \------>

Note that If T1-init timer expires at "A" after the INIT or COOKIE
chunks are sent, the same INIT or cookie chunk with the same Initiate
Tag (i.e., Tag_A) or cookie shall be retransmitted and the timer
restarted. This shall be repeated Max.Init.Retransmit times before "A"
considers "Z" unreachable and reports the failure to its upper layer.
When retransmitting the INIT, the endpoint SHALL following the rules
defined in 5.3 to determine the proper timer value.

4.2 Handle Duplicate INIT, INIT ACK, COOKIE, and COOKIE ACK

At any time during the life of an association (in one of the possible
states) between an endpoint and its peer, one of the setup chunks
may be received from the peer, the receiver shall process such
a duplicate has described in this section.

The following scenarios can cause duplicated chunks:

A) The peer has crashed without being detected, and re-started itself
   and sent out a new Chunk trying to restore the association,

B) Both sides are trying to initialize the association at about the
   same time,

C) The chunk is a staled datagram that was used to establish
   the present association or a past association which is no longer in
   existence, or

D) The chunk is a false message generated by an attacker.

In case A), the endpoint shall reset the present association and set a
new association with its peer. Case B) is unique and is discussed in
Section 4.2.1. However, in cases C) and D), the endpoint must retain
the present association.

The rules in the following sections shall be applied in order to
identify and correctly handle these cases.

4.2.1 Handle Duplicate INIT in COOKIE-WAIT or COOKIE-SENT State

This usually indicates an initialization collision.

In such a case, each of the two side shall respond to the other side
with an INIT ACK, with the Verification Tag field of the common header
set to the tag value received from the INIT message, and the Initiate
Tag field set to its own tag value (the same tag used in the INIT
message sent out by itself). Each responder shall also generate a
cookie with the INIT ACK.

After that, no other actions shall be taken by either side, i.e., the
endpoint shall not change its state, and the T1-init timer shall be
let running. The normal procedures for handling cookies will
resolve the duplicate INITs to a single association.

4.2.2 Handle Duplicate INIT in Other States

Upon reception of the duplicated INIT, the receiver shall follow
the normal procedures for handling a INIT message, i.e. generate
a INIT ACK with a cookie.

In the outbound INIT ACK, the Verification Tag field of the common
header shall be set to the peer tag value (from the INIT message), and
the Initiate Tag field set to its own tag value (unchanged from the
existing association). A cookie should also be included generated
with the current time and a updated TCB based upon the INIT message.
 And no further actions shall be taken.

4.2.3 Handle Duplicate INIT ACK

If an INIT ACK is received by an endpoint in any state
other than the COOKIE-WAIT state, the endpoint should discard
the INIT ACK message. A duplicate INIT ACK usually indicates the
processing of a old INIT or duplicated INIT message.

4.2.4 Handle Duplicate Cookie

When a duplicated COOKIE chunk is received in any state for an
existing association the following rules shall be applied:

1) compute an MD5 signature using the TCB data carried in the cookie
   along with the receiver's private security key,

2) authenticate the cookie by comparing the computed MD5 signature
   against the one carried in the cookie. If this comparison fails,
   the datagram, including the COOKIE and the attached user data,
   should be silently discarded (this is case C or D above).

3) compare the timestamp in the cookie to the current time, if
   the cookie is older than the lifespan carried in the cookie,
   the datagram, including the COOKIE and the attached user data,
   should be discarded and the endpoint MUST transmit a stale cookie
   error to the sending endpoint only if the Verification tags of the
   cookie's TCB does NOT match the current tag values in the association
   (this is case C or D above).

4) If the cookie proves to be valid, unpack the TCB into a
   temporary TCB.

5) If the Verification Tags in the Temporary TCB matches the
   Verification Tags in the existing TCB, the cookie is a
   duplicate cookie. A cookie ack should be sent to the peer
   endpoint but NO update should be made to the existing
   TCB.

6) If the the local Verification Tag in the temporary TCB
   does not match the local Verification Tag in the existing
   TCB, then the cookie is a old stale cookie and does
   not correspond to the existing association (case C above).
   The datagram should be silently discarded.

7) If the Peers Verification Tag in the temporary TCB does not
   match the Peers Verification Tag in the existing TCB
   then a restart of the peer has occurred (case A above) and the
   endpoint should report the restart and respond with a COOKIE-ACK
   message; Updating the Verification Tag, starting sequence
   number, and network information of its peer from the temporary
   TCB to the existing TCB. After which the temporary TCB may be
   discarded.

  IMPLEMENTATION NOTE: It is an implementation decision on how
  to handle any pending datagrams. The implementation may elect
  to either A) send all messages up to its upper layer with the
  restart report, or B) automatically requeue any datagrams
  pending, marking all to the unsent state and assigning
  new TSN's at the time of initial transmit based upon the
  updated starting sequence number (as defined in section 5.5).

4.2.5 Handle Duplicate COOKIE-ACK.

At any state other than COOKIE-Sent a endpoint may receive a
duplicated COOKIE-ACK chunk. If so, the chunk should be silently
discarded.

4.2.6 Handle Stale COOKIE Error

A stale cookie error indicates one of a number of possible events:

A) that the association failed to completely setup before the
   cookie issued by the sender was processed.

B) an old cookie was processed after setup completed.

C) an old cookie is received from someone that the receiver is
   not interested in having a association with and the ABORT
   message was lost.

When processing a stale cookie a endpoint should first examine
if an association is in the process of being setup, i.e. the
association is in the COOKIE-SENT state. In all cases if
the association is NOT in the COOKIE-SENT state, the stale
cookie message should be silently discarded.

If the association is in the COOKIE-SENT state, the endpoint
may elect one of three alternatives.

1) Send a new INIT message to the endpoint, to generate
   a new cookie and re-attempt the setup procedure.

2) Discard the TCB and report to the upper layer the
   inability to setup the association.

3) Send a new INIT message to the endpoint, adding a cookie
   preservative requesting a time extentsion on the life of the
   cookie. When calculating the time extension, an implementation
   SHOULD use Round Trip Time (RTT) information generated from
   the original COOKIE <-> Stale COOKIE timing and should add no more
   than 1,000,000 microseconds beyond the RTT. Long cookie lives will
   make a endpoint more subject to a replay attack.

4.3 Other Initialization Issues

4.3.1 Selection of Tag Value

Initiate Tag values should be selected from the range of 0x1 to
0xffffffff. It is very important that the Tag value be randomized to
help protect against "man in the middle" and "sequence number" attacks.
It is suggested that RFC 1750 [1] be used for the Tag randomization.

Moreover, the tag value used by either endpoint in a given association
MUST never be changed during the lifetime of the association. However,
a new tag value MUST be used each time the endpoint tears-down and
then re-establishes the association to the same peer.

4.3.2 Initiation from behind a NAT

When a NAT is present between two endpoints, the endpoint that is
behind the NAT, i.e., one that does not have a publicly available
network address, shall take one of the following options:

A) Indicate that only one address can be used by including no transport
addresses in the INIT message (Section 2.3.1.1). This will make the
endpoint that receives this Initiation message to consider the sender
as only having that one address. This method can be used for a dynamic
NAT, but any multi-homing configuration at the endpoint that is behind
the NAT will not be visible to its peer, and thus not be taken
advantage of.

B) Indicate all of its networks in the Initiation by specifying all
the actual IP addresses and ports that the NAT will substitute for the
endpoint. This method requires that the endpoint behind the NAT must
have pre-knowledge of all the IP addresses and ports that the NAT will
assign.

5. User Data Transfer

For transmission efficiency, SCTP defines mechanisms for bundling of
small user messages and segmentation of large user messages.

The following diagram depicts the flow of user messages through SCTP.

              +--------------------------+
              |      User Messages       |
              +--------------------------+
    SCTP user        ^  |
   ==================|==|=======================================
                     |  v (1)
          +------------------+    +--------------------+
          | SCTP DATA Chunks |    |SCTP Control Chunks |
          +------------------+    +--------------------+
                     ^  |             ^  |
                     |  v (2)         |  v (2)
                  +--------------------------+
                  |      SCTP datagrams      |
                  +--------------------------+
    SCTP                      ^  |
   ===========================|==|===========================
                              |  v
    Unreliable datagram service (e.g., UDP)

   Note:
   (1) When converting user messages into Data chunks, SCTP sender
       will segment user messages larger than the current path MTU
       into multiple data chunks. The segmented message will be
       reassembled from data chunks before delivery by the SCTP
       receiver.

   (2) Multiple data and control chunks may be multiplexed by the
       sender into a single SCTP datagram for transmission, as long as
       the final size of the datagram does not exceed the current path
       MTU. The receiver will de-multiplex the datagram back into
       chunks.

The bundling and segmentation mechanisms, as detailed in Sections 5.9
and 5.10, are optional to implement by the data sender, but they MUST
be implemented by the data receiver, i.e., a SCTP receiver MUST be
prepared to receive and process bundled or segmented data.

5.1  Transmission of DATA Chunks

The following general rules SHALL be applied by the sender for
transmission and/or retransmission of outbound DATA chunks:

A) At any given time, the sender MUST NOT transmit new data onto any
   destination transport address if it has rwnd or more octets of data
   outstanding. The outstanding data size is defined as the total size
   of ALL data chunks outstanding.

B) At any given time, the sender MUST NOT transmit new data onto a
   given transport address if it has cwnd or more octets of data
   outstanding on that transport address.

C) When the time comes for the sender to transmit, before sending
   new DATA chunks, the sender MUST first transmit any outstanding
   DATA chunks which are marked for retransmission.

D) Then, the sender can send out as many new DATA chunks as Rule A and
   Rule B above allow.

Note: multiple DATA chunks committed for transmission MAY be
bundled in a single packet, unless bundling is explicitly disallowed
by ULP of the data sender. Also, it is implementation specific
whether to allow the bundling of retransmission DATA chunks with
new DATA chunks.

Note: if there are any unacknowledged DATA chunks (e.g., due to
delayed ack), the sender should create a SACK and bundle it with
the outbound DATA chunk, as long as the size of the final SCTP
datagram does not exceed the current MTU. See Section 5.2.

  IMPLEMENTATION Note: if the window is full (i.e., transmission is
  disallowed by Rule A and/or Rule B), the sender MAY still accept
  send requests from its upper layer, but SHALL transmit no more DATA
  chunks until some or all of the outstanding DATA chunks are
  acknowledged and transmission is allowed by Rule A and Rule B
  again.

In all cases, if a transmission or retransmission of a DATA chunk is
made, the sender shall start the retransmission timer (T3-rxt) if it
is not currently running. The T3-rxt timer value must be adjusted
according to the timer rules defined in Sections 5.3.2, and 5.3.3.

Note: If rwnd is set to 0, and all outstanding DATA chunks are
acknowledged, a sender is allowed to send ONE new DATA chunk up to the
size of current path MTU, to probe the receiver for changes to rwnd.

5.2  Acknowledgment on Reception of DATA Chunks

The SCTP receiver MUST always acknowledge the SCTP sender about the
reception of each DATA chunk.

The guidelines on delayed acknowledgment algorithm specified in
Section 4.2 of RFC 2581 [3] SHOULD be followed. In particular, a SCTP
receiver MUST NOT excessively delay acknowledgments. Specifically, an
acknowledgement SHOULD be generated for at least every second datagram
received, and SHOULD be generated within 200 ms of the arrival of any
unacknowledged datagram.

  IMPLEMENTATION NOTE: the maximal delay for generating an
  acknowledgement may need to be configurable by the SCTP user,
  either statically or dynamically, in order to meet the specific
  timing requirement of the signaling protocol being carried.

Acknowledgments MUST be sent in SACK control chunks. A SACK chunk can
acknowledge the reception of multiple DATA chunks. See Section 2.3.3
for SACK chunk format. In particular, the SCTP receiver MUST fill the
Highest Consecutive TSN ACK field to indicate the highest consecutive
TSN number it has received, and any received segments beyond the
highest consecutive TSN SHALL also be reported.

Upon reception of the SACK, the data sender MUST adjust its total
outstanding data count and the outstanding data count on those
destination addresses for which one or more data chunks is
acknowledged by the SACK.

The following example illustrates the use of delayed acknowledgments:

Endpoint A                                      Endpoint Z

{App sends 3 messages; strm 0}
DATA [TSN=7,Strm=0,Seq=3] ------------> (ack delayed)
(Start T3-rxt timer)

DATA [TSN=8,Strm=0,Seq=4] ------------> (send ack)
                              /------- SACK [TSN ACK=8,Frag=0]
(cancel T3-rxt timer)  <-----/
...
...

DATA [TSN=9,Strm=0,Seq=5] ------------> (ack delayed)
(Start T3-rxt timer)
                                       ...
                                       {App sends 1 message; strm 1}
                                       (bundle SACK with DATA)
                                /----- SACK [TSN Ack=9,Seg=0] \
                               /         DATA [TSN=6,Strm=1,Seq=2]
(cancel T3-rxt timer)  <------/        (Start T3-rxt timer)
(ack delayed)
...
(send ack)
SACK [TSN ACK=6,Seg=0] --------------> (cancel T3-rxt timer)

5.3 Management of Retransmission Timer

SCTP uses a retransmission timer T3-rxt to ensure data delivery in the
absence of any feedback from the remote data receiver. The duration of
this timer is referred to as RTO (retransmission timeout).

The computation and management of RTO in SCTP follows closely with how
TCP manages its retransmission timer. To compute the current RTO, an
SCTP sender maintains two state variables per destination transport
address: SRTT (smoothed round-trip time) and RTTVAR (round-trip time
variation).

5.3.1 RTO Calculation

The rules governing the computation of SRTT, RTTVAR, and RTO are
as follows:

C1) Until an RTT measurement has been made for a packet sent
    to the given destination transport address, set RTO to the
    protocol parameter 'RTO.Initial'.

C2) When the first RTT measurement R is made, set SRTT <- R,
    RTTVAR <- R/2, and RTO <- SRTT + K * RTTVAR, where K = 4.

C3) When a new RTT measurement R' is made, set

	RTTVAR <- beta * RTTVAR + (1 - beta) * |SRTT - R'|
	SRTT <- alpha * SRTT + (1 - alpha) * R'

    (The value of SRTT used in the update to RTTVAR is its value
    *before* updating SRTT itself using the second assignment.)

    The above are computed using alpha=1/8 and beta=1/4.

    After the computation, update RTO <- SRTT + 4 * RTTVAR.

C4) It is RECOMMENDED that new RTT measurements should be made
    no more than once per round-trip for a given destination
    transport address. There are two reasons for this recommendation:
    first, it appears that measuring more often does not in practice
    yield any significant benefit [5]; second, if measurements
    are made more often, then the values of alpha and beta in
    rule C3 above must be adjusted so that SRTT and RTTVAR still
    adjust to changes at roughly the same rate (in terms of
    how many round trips it takes them to reflect new value) as
    they would if making only one measurement per round-trip
    and using alpha and beta as given in rule C3.

C5) Karn's algorithm: RTT measurements MUST NOT be made using
    packets that were retransmitted (and thus for which it is
    ambiguous whether the reply was for the first instance of the
    packet or a later instance).

C6) Whenever RTO is computed, if it is less than 1 second then
    it is rounded up to 1 second.  The reason for this rule is
    that RTOs that do not have a high minimum value are susceptible
    to unnecessary timeouts [5].

C7) A maximum value may be placed on RTO provided it is at least
    60 seconds.

There is no requirement for the clock granularity G used for computing
RTT measurements and the different state variables, other than

G1) Whenever RTTVAR is computed, if RTTVAR = 0, then adjust
    RTTVAR <- G.

Experience has shown that finer clock granularities (<= 100 msec)
perform somewhat better than more coarse granularities.

5.3.2 Retransmission Timer Rules

The rules for managing the retransmission timer are as follows:

R1) Every time a packet containing data is sent (including a
    retransmission), if the T3-rxt timer is not running, start it
    running so that it will expire after RTO seconds. The RTO
    used here is that obtained after any doubling due to
    retransmission as discussed in rule E3 below.

R2) Whenever all outstanding data has been acknowledged, turn
    off the retransmission timer.

R3) Whenever a SACK is received that acknowledges new data chunks
    including the one with the earliest outstanding TSN (i.e., moving
    the cumulative ACK point forward),
    restart T3-rxt retransmission timer so that it will expire
    after RTO seconds (for the current value of RTO).

The following example shows the use of various timer rules (assuming
the receiver uses delayed acks).

Endpoint A                                         Endpoint Z
{App sends 2 messages; strm 0}
Data [TSN=7,Strm=0,Seq=3] ------------> (ack delayed)
(Start T3-rxt timer)
                                        {App sends 1 message; strm 1}
                                        (bundle ack with data)
DATA [TSN=8,Strm=0,Seq=4] ----\     /-- SACK [TSN ACK=7,Frag=0] \
                               \   /      DATA [TSN=6,Strm=1,Seq=2]
                                \ /     (Start T3-rxt timer)
                                 \
                                / \
(Re-start T3-rxt timer) <------/   \--> (ack delayed)
(ack delayed)
...
{send ack}
SACK [TSN ACK=6,Frag=0] --------------> (Cancel T3-rxt timer)
                                        ..
                                        (send ack)
(Cancel T3-rxt timer)  <-------------- SACK [TSN ACK=8,Frag=0]

5.3.3 Handle T3-rxt Expiration

Whenever the retransmission timer T3-rxt expires, do the following:

E1) Adjust ssthresh with rules defined in Section 6.2.3.

E2) Determine how many of the earliest (i.e., lowest TSN)
    outstanding Data chunks will fit into a single packet,
    subject to the MTU constraints for the path corresponding to
    the destination transport address. Call this value K.
    Retransmit those K data chunks in a single packet, and set
    cwnd <- MTU.

E3) Set RTO <- RTO * 2 ("back off the timer"). The maximum value
    discussed in rule C7 above may be used to provide an upper bound
    to this doubling operation.

E4) Start the retransmission timer, per rule R1 above.

    Note, the data sender MAY use a value smaller than the RTO when
    start the retransmission timer IF the sender has fewer than
    cwnd octets of outstanding data on the transport address to which
    the retransmission is being sent.

    [EditorĘs

    [Editors Note: if the receiver is multi-homed and the retrans
     is to be sent to an alternate address, how this rapid retran
     rule should apply??]

Note that after retransmitting, once a new RTT measurement is obtained
(which can only happen when new data has been sent and acknowledged,
per rule C5, or for a measurement made from a Heartbeat [see Section
7.3]), the computation in rule C3 is performed, including the
computation of RTO, which may result in "collapsing" RTO back down
after it has been subject to doubling (rule E3).

The final rule for managing the retransmission timer concerns failover:

F1) Whenever SCTP switches from the current destination transport
    address to a different one (per section ???), 5.4), the current
    retransmission timer is left running.  As soon as SCTP transmits
    a packet containing data to the new transport address, restart the
    timer, using the RTO value for the path to the new address.

5.4 Multi-homed SCTP Endpoints

An SCTP endpoint is considered multi-homed if there are more than one
transport addresses that can be used as a destination address to reach
that endpoint.

Moreover, at the sender side, one of the multiple destination
addresses of the multi-homed receiver endpoint shall be selected as
the primary destination transport address by the UPL (see Section 9
for details).

At association initiation, the initial primary destination transport
addresses are:

- for the sender of the INIT message, the transport address that the
  INIT is sent on. This may be changed upon reception of the
  destination transport address list in the INIT ACK message from the
  peer.

- for the sender of the INTI ACK message, any valid transport address
  obtained from the INIT message.

When the SCTP sender is transmitting to the multi-homed receiver, by
default the transmission SHOULD always take place on the primary
transport address, unless the SCTP user explicitly specifies the
destination transport address to use.

If possible, acknowledgements SHOULD be transmitted to the same
destination transport address from which the acknowledged DATA or
control chunks were received (Note: when acknowledging multiple DATA
chunks in a single SACK, this may not be possible).

Some of the destination transport addresses may become inactive due to
either the occurrance of certain error conditions or adjustments from
SCTP user.

In the case where the primary destination transport address becomes
inactive, or the SCTP user tries to explicitly send to an inactive
destination transport address, the SCTP sender should either send the
chunk to an alternate active destination transport address, or to
report an error. This is implementation specific.

Also, when the SCTP receiver is multi-homed, an SCTP sender SHOULD
always try to retransmit a chunk to an active destination transport
address that is different from the original destination address used
to transmit that chunk.  Retransmissions do not affect the total
outstanding data count. However, if the data chunk is retransmitted
onto a different destination address, the outstanding data counts on
the new destination address and the old destination address where the
data chunk was originally sent to shall be adjusted accordingly.

5.5 Stream Identifier and Sequence Number

Every DATA chunk MUST carry a valid stream identifier. If a DATA chunk
with an invalid stream identifier is received, the receiver shall
respond immediately with an ERROR message with cause set to Invalid
Stream Identifier (see Section 2.3.9) and discard the DATA chunk.

The stream sequence number in all the streams shall start from 0x0
when the association is established. Also, when the stream sequence
number reaches the value 0xffff the next sequence number shall be set
to 0x0.

5.6 Ordered and Un-ordered Delivery

Normally, the SCTP receiver shall ensure the DATA chunks within any
given stream be delivered to the upper layer according to the order of
their stream sequence number. If there are DATA chunks arriving out of
order of their stream sequence number, the receiver MUST hold the
received DATA chunks from delivery until they are re-ordered.

However, an SCTP sender can indicate that no ordered delivery is
required on a particular DATA chunk within the stream by setting the U
flag of the DATA chunk to 1.

In this case, the receiver must ignore the sequence number field of
the data chunk, bypass the ordering mechanism and immediately delivery
the data to the upper layer (after re-assembly if the user data is
segmented by the sender).

This provides an effective way of transmitting "out-of-band" data in a
given stream. Also, a stream can be used as an "unordered" stream by
simply setting the U flag to 1 in each outbound DATA chunk from that
stream.

  IMPLEMENTATION NOTE: An implementation, when sending an unordered
  DATA chunk, may choose to place the DATA chunk in an outbound
  datagram at the head of the outbound transmission queue if
  possible.

5.7 Report Gaps in Received DATA TSNs

Upon the reception of a new DATA chunk, an SCTP receiver shall examine
the continuity of the TSNs received. If the receiver detects that gaps
exist in the received DATA chunk sequence, an SACK with fragment
reports shall be sent back immediately.

Based on the segment reports from the SACK, the data sender can
calculate the missing DATA chunks and make decisions on whether to
retransmit them (see Section 5.3 for details).

Multiple gaps can be reported in one single SACK (see Section 2.3.3).

Note that when the data sender is multi-homed, the SCTP receiver
SHOULD always try to send the SACK to the same network from where the
last DATA chunk was received.

Upon the reception of the SACK, the data sender SHALL remove all DATA
chunks which have been acknowledged by the SACK. The data sender MUST
also treat all the DATA chunks which fall into the gaps between the
fragments reported by the SACK as "missing". The number of "missing"
reports for each outstanding DATA chunk MUST be recorded by the data
sender in order to make retransmission decision, see Section 6.2.4 for
details.

The following example shows the use of SACK to report a gap.

Endpoint A                                    Endpoint Z
{App sends 3 messages; strm 0}
DATA [TSN=6,Strm=0,Seq=2] ---------------> (ack delayed)
(Start T3-rxt timer)

DATA [TSN=7,Strm=0,Seq=3] --------> X (lost)

DATA [TSN=8,Strm=0,Seq=4] ---------------> (gap detected,
                                            immediately send ack)
                                /----- SACK [TSN ACK=6,Frag=1,
                               /             Strt=2,End=2]
                        <-----/
(remove 6 and 8 from out-queue,
 and strike 7 as "1" missing report)

Note: in order to keep the size of the outbound SCTP datagram not to
exceed the current path MTU, the maximal number of fragments that can
be reported within a single SACK chunk is limited. When a single SACK
can not cover all the fragments needed to be reported due to the MTU
limitation, the endpoint SHALL send only one SACK, reporting the
fragments from the lowest to highest TSNs, within the size limit set
by the MTU, and leave the remaining highested TSN fragment numbers
unacknowledged.

5.8 CRC-16 Utilization

When sending a datagram, the sender can choose to strengthen the data
integrity of the transmission by including the CRC-16 value calculated
on the datagram, as described below.

After the datagram is constructed (containing the SCTP common header
and one or more control or DATA chunks), the sender shall:

1) fill in the proper Version number and Verification Tag in the
   common header,

2) set the C Bit to '1' and fill the 16 bit CRC-16 field with '0',

3) calculate the CRC-16 value of the whole datagram, including the
   SCTP common header and all the chunks,

   It shall be the ones complement of the sum (modulo 2) of:
   a) the remainder of x k (x 15 + x 14 + x 13 + x 12 + x 11 + x 10 +
      x 9 + x 8 + x 7 + x 6 + x 5 + x 4 + x 3 + x 2 + x + 1) divided
      (modulo 2) by the generator polynomial x 16 + x 12 + x 5 + 1,
      where k is the number of bits in the SCTP frame not including
      the CRC-16 bits, and
   b) the remainder of the division (modulo 2) by the generator
      polynomial x 16 + x 12 + x 5 + 1, of the product of x 16 by the
      content of the frame not including the CRC-16 bits.

4) put the resultant value into the CRC-16 field, and leave the rest of
   the bits unchanged.

When a datagram is received, the receiver MUST first check the C
Bit. If the C Bit is set, the receiver SHALL:

1) store the received CRC-16 value aside,

2) replace the 16 bits of the CRC-16 with '0' and calculate a CRC-16
   value of the whole received datagram,

3) verify that the calculated CRC-16 value is the same as the received
   CRC-16 value, If not, the receiver MUST treat the datagram as an
   invalid SCTP datagram.

If the C Bit is not set, the receiver MUST NOT perform the above
CRC-16 check.

The default procedure of handling invalid SCTP datagrams is to
silently discard them.

5.9 Segmentation

Segmentation SHALL be performed by the data sender if the user message
to be sent has a large size that causes the outbound SCTP datagram
size exceeding the current MTU.

When determining when to segment, the SCTP implementation MUST take
into account the SCTP datagram header as well as the DATA chunk
header. The implementation MAY also take account of the space required
for a SACK chunk.

  IMPLEMENTATION NOTE: if the data receiver is multi-homed, the
  latest MTU of the current primary destination address shall be
  used.

  IMPLEMENTATION NOTE: if segmentation is not support by the sender,
  an error should be reported to the sender's SCTP user if the data to be
  sent has a size exceeding the current MTU. In such cases the Send
  primitive discussed in Section 9.1 would need to return an error
  to the upper layer.

Segmentation takes the following steps:

1) the data sender SHALL break the large user message into a series of
   DATA chunks, each with a total size smaller than the current MTU
   minus the SCTP common header size (i.e., 8 octets),

2) the data sender MUST then assign, in sequence, a separate TSN to
   each of the DATA chunks in the series,

3) the data sender MUST also set the B/E bits of the first DATA chunk
   in the series to '10', the B/E bits of the last DATA chunk in the
   series to '01', and the B/E bits of all other DATA chunks in the
   series to '00'.

The data receiver MUST recognize the segmented DATA chunks, by
examining the B/E bits in each of the received DATA chunks, and queue
the segmented DATA chunks for re-assembly. Then, it shall pass the
re-assembled user message to the specific stream for re-ordering and
final dispatching.

5.10 Bundling and Multiplexing

An SCTP sender achieves data bundling by simply including multiple
DATA chunks in one outbound SCTP datagram. Note that the total size of
the resultant SCTP datagram, including the SCTP common header, MUST be
less or equal to the current MTU.

  IMPLEMENTATION NOTE: if the data receiver is multi-homed, the
  latest MTU of the current primary destination address shall be
  used.

When multiplexing control chunks with DATA chunks, control chunks have
the priority and MUST be placed first in the outbound SCTP datagram
and be transmitted first. The transmitter MUST transmit DATA chunks
within a SCTP datagram in increasing order of TSN.

Partial chunks MUST NOT be placed in a SCTP datagram.

The receiver MUST process the chunks in order in the datagram.  The
receiver uses the chunk length field to determine the end of a chunk
and beginning of the next chunk taking account of the fact that all
chunks end on a thirty-two-bit word boundary. If the receiver detects
a partial chunk, it MUST drop the chunk.

6. Congestion control

[ Editors Notes.. this section is still being reworked and may
  have some changes ]

Congestion control is one of the basic functions in the SCTP protocol.
It is likely that adequate resources will be allocated to SCTP traffic
to assure prompt delivery of time-critical SCTP data, thus it would be
unlikely, during normal operations, that SCTP transmissions encounter
severe congestion condition.  However SCTP must prepare itself for
adverse operational conditions, which can develop upon partial network
failures or unexpected traffic surge.  In such situations SCTP must
follow correct congestion control steps to recover from congestion
quickly in order to get data delivered as soon as possible.  In the
absence of network congestion, these preventive congestion control
algorithms will show no impact on the protocol performance.

The congestion control algorithms used by SCTP are based on RFC 2581
[3], "TCP Congestion Control".  This section describes how the
algorithms defined in RFC 2581 are adopted for use in SCTP.  We first
list differences in protocol designs between TCP and SCTP, and then
describe SCTP's congestion control scheme.  The description will use
the same terminology as in TCP congestion control whenever
appropriate.

6.1 SCTP Differences from TCP Congestion control

One difference from TCP is that Selective Acknowledgment function
(SACK) is designed into SCTP, rather than an enhancement that is added
to the protocol later as the case for TCP.  SCTP SACK carries
different semantic meanings from that of TCP SACK.  TCP considers the
information carried in the SACK as advisory information only.  In
SCTP, any DATA chunk that has been acknowledged by SACK, including
DATA that arrived at the receiving end out of order, are considered
having been delivered to the destination application, and the sender
is free to discard the local copy.  Thus the value of cwnd controls
the number of outstanding data; it is not the upper bound between the
highest acknowledged sequence number and the latest DATA chunk that
can be sent within the congestion window, as the case in TCP.  SCTP
SACK leads to different implementations of fast-retransmit and
fast-recovery from that of TCP.

The biggest difference between SCTP and TCP, however, is multi-homing.
SCTP is designed to establish robust communication associations
between two end points each of which may be reachable by more than one
transport address.  Potentially different addresses may lead to
distinguished data paths between the two points, thus ideally one may
need a separate set of congestion control parameters for each of the
paths.  Given SCTP is the first transport protocol whose design
specifically takes multihoming issue into consideration, however,
there is no experience at this point regarding the use of multiple
addresses, or the likelyhood of each address pair representing a
separate path.  To proceed with caution, we make the following
assumptions:

o the sender does not change the use of source address often, if at
  all.
o the sender always uses the same destination address until being
  instructed by the upper layer otherwise.
o the sender keeps a separate congestion control parameter set for each
  of the destination addresses. The parameters should decay if the
  address is not used for a long enough time period.
o For each of the destination addresses, do slow-start upon the first
  transmission to that address.

6.2 SCTP Slow-Start and Congestion Avoidance

The slow start and congestion avoidance algorithms MUST be used by a
SCTP sender to control the amount of outstanding data being injected
into the network.  The congestion control in SCTP is employed in regard
to the association, not to an individual stream.  In some situations it
may be beneficial for a SCTP sender to be more conservative than the
algorithms allow, however a SCTP sender MUST NOT be more aggressive than
the following algorithms allow.

Like TCP, an SCTP sender uses the following three control variables to
regulate its transmission rate.

o Receiver advertised window size (rwnd), which is set by the
  receiver based on its available buffer space for incoming packets.
o Congestion control window (cwnd), which is adjusted by the sender
  based on observed network conditions.
o Slow-start threshold (ssthresh), which is also used by the sender to
  distinguish congestion control and congestion avoidance phases.

SCTP also requires one additional control variable, cwnd2, which is used
during congestion avoidance phase to facilitate cwnd adjustment.

6.2.1 Slow-Start

Beginning data transmission into a network with unknown conditions
requires SCTP to probe the network to determine the available capacity.
The slow start algorithm is used for this purpose at the beginning of a
transfer, or after repairing loss detected by the retransmission timer.

o The initial value of cwnd MUST be less than or equal to 2*MTU octets.
o The initial value of ssthresh MAY be arbitrarily high (for example,
  some implementations use the size of the receiver advertised window).
o Whenever cwnd is greater than zero, the sender is allowed to have cwnd
  octets of data outstanding on that transport address.
o When cwnd is less than or equal to ssthresh AND the sender has cwnd or more
  data outstanding on the transport address, cwnd is incremented by
  the total number of octets of all new data chunks acknowledged in each
  SACK received (piggy-backed or stand-alone SACKs).
o When the sender does not transmit data on a given transport address, the
  cwnd of the transport address should be adjusted to
  max(cwnd / 2, 2*MTU) per RTT.

Note: a piggy-backed SACK is one that is bundled with a DATA chunk.

6.2.2 Congestion Avoidance

Whenever cwnd is increased to be equal or greater than ssthresh, cwnd
should be incremented by MTU per RTT if at time the sender has cwnd or
more octets of data outstanding on that transport address.

In practice an implementation can
achieve this goal in the following way:

o cwnd2 is initialized to 0.
o Whenever cwnd is equal or greater than ssthresh, upon each SACK
  arrival, increase cwnd2 by the total number of octets of all new
  chunks acknowledged in that SACK.
o When cwnd2 is equal or greater than cwnd and before the arrival of the
  SACK the sender has cwnd or more octets of data outstanding,
  increase cwnd by MTU, and reset cwnd2 to (cwnd2 - cwnd).

6.2.3 Congestion Control

Upon detection of packet losses from SACK, the sender should do the
following:

  ssthresh = max(cwnd/2, 2*MTU)
  cwnd = ssthresh/2

Basically, a packet loss causes cwnd to be cut in half(or 1/4??).

When the T3-rxt timer expires, SCTP should perform slow start by
setting cwnd = MTU, and assure that no more than one DATA chunk will
be in flight until the sender receives acknowledgment for successful
delivery.

6.2.4 Fast Retransmit on Gap Reports

In the absence of data losses, a SCTP receiver performs delayed
acknowledgment.  However whenever a receiver notices a hole in the
arriving TSN sequence, it should start sending a SACK for every
packet arrival.

At the sender end, whenever the sender notices a hole in a SACK, it
should wait for 3 further SACKs before taking action.  If the 3
subsequent SACKs report the same TSN(s) missing, the sender shall:

1) mark the DATA chunk(s) for retransmission,
2) adjust its ssthresh and cwnd according to the formula described in
   Section 6.2.3.
3) Restart T3-rxt timer if it is running, and
4) start retransmission procedure, as described in Section 5.3.3.

A straightforward implementation of the above requires that the sender
keeps a counter for each TSN hole first reported by a SACK; the
counter keeps track of whether 3 subsequent SACKs have reported the
same hole.

Because cwnd in SCTP bounds the number of outstanding TSN's,
the effect of TCP fast-recovery is achieved automatically with no
adjustment to the control window size. [Is this still true??]

6.3 Path MTU Discovery

    [Editor's Note: text

RFC 1191 discusses "Path MTU Discovery", whereby a sender maintains an
estimate of the maximum transmission unit (MTU) along a given Internet path
and refrains from sending datagrams along that path which exceed the MTU,
other than occasional attempts to probe for a change in the path MTU.
RFC 1191 is thorough in its discussion of the MTU discovery mechanism and
strategies for determining the current end-to-end MTU setting as well as
detecting changes in this value.  RFC 1981 discusses applying the same
mechanisms for IPv6.

An SCTP sender MUST apply these techniques, and MUST do so on a
per-destination-address basis.

There are 4 ways in which SCTP differs from the description in RFC 1191 of
applying MTU discovery to TCP:

	1)  SCTP associations can span multiple set of addresses.
	    Per the above comment, an SCTP sender MUST maintain separate
	    MTU estimates for each destination address of its peer.

	2)  Elsewhere in this document, when the term "MTU" is discussed,
	    it refers to the MTU associated with the destination address
	    corresponding to the context of the discussion.  If this
	    address is ambiguous, it indicates a bug in this document.

	3)  Unlike TCP, SCTP does not have a notion of "Maximum Segment
	    Size".  Accordingly, the MTU for each destination address
	    SHOULD be provided by Vern] initialized to a value no larger than the link MTU
	    for the local interface to which datagrams for that remote
	    destination address will be routed.

	4)  Since data transmission in SCTP is naturally structured in
	    terms of TSNs rather than bytes (as is the case for TCP), the
	    discussion in section 6.5 of RFC 1191 applies: when retransmitting
	    a datagram to a remote address for which the datagram appears
	    too large for the path MTU to that address, the datagram SHOULD
	    be retransmitted without the DF bit set, allowing it to possibly
	    be fragmented.  Transmissions of new datagrams MUST have DF set.

Other than these differences, the discussion of TCP's use of MTU discovery
in RFCs 1191 and 1981 applies to SCTP, too, on a per-destination-address
basis.

6.4 Discussion

[The following discussion needs to be updated for byte-based algorithm]

There is one important difference between the SCTP congestion control,
as described above, and TCP congestion control.  In the latter the
control behavior is measured in unit of packets.  That is, upon slow
start, a TCP connection doubles cwnd value per RTT as measured by the
number of packets.  In SCTP, cwnd value is doubled per RTT as measured
by the number of DATA chunks.  Similarly, during congestion avoidance,
in the absence of packet losses a TCP connection increases cwnd by one
data packet per RTT, while a SCTP association increases cwnd by one DATA
chunk per RTT.

Ideally, congestion control should be performed by controlling the
number of outstanding packets.  However because the DATA chunk size is a
variable, there does not seem a simple and reliable way to translate "one
datagram" to a suitable number of chunks.  Although the SCTP congestion
control design, as described in this section, represents a simple
starting point, it may have potential negative impact on the application
performance depending on the relative sizes of DATA chunks and packet
MTU.  For example, if the MTU is 5 times of the average DATA chunk size,
it would take 5 times longer for a SCTP association to open up its cwnd
to the same size of that of a TCP connection.  During a slow start
congestion recovery, limiting the transmission to cwnd DATA chunks may
lead to sending half-full packets, even when more application data is
waiting to be transmitted.

We suggest that the current design be discussed, and revised if deemed
necessary.

7.  Fault Management

7.1 Endpoint Failure Detection

The data sender shall keep a counter on the total number of
consecutive retransmissions to its peer (including retransmissions to
ALL the destination transport addresses of the peer if it is
multi-homed).

If the value of this counter exceeds the limit defined in the protocol
parameter 'Max.Retransmits', the data sender shall consider the peer
endpoint unreachable and shall stop transmitting any more data to
it. In addition, the data sender shall report the failure to the upper
layer, and optionally report back all outstanding datagrams remaining
in its outbound queue.

The counter shall be reset each time a datagram is received from the
peer endpoint.

7.2 Path Failure Detection

When the remote endpoint is multi-homed, the data sender should keep a
'retrans.count' counter for each of the destination transport addresses of
the remote endpoint.

This count should be incremented each time the data sender retransmits
an outstanding datagram which was originally sent to the destination
transport address.

When the value in 'retrans.count' exceeds half of the value of the
protocol parameter 'Max.Retransmits', the data sender should mark the
corresponding destination transport address as inactive, and a notification
may optionally be sent to the upper layer.

When an outstanding datagram is acknowledged, the data sender should
clear the 'retrans.count' counter of the destination transport address to
which the datagram was sent. In the case of a retransmitted datagram
(due to time-out or SACK) the destination transport address last
sent to should be used to determine which 'retrans.count' to clear.

7.3 Path Heartbeat

By default, an SCTP endpoint shall monitor the reachability of the
idle destination transport address(es) of its peer by sending
HEARTBEAT messages periodically to the destination transport
address(es).

A destination transport address should be considered idle if no
datagram has been sent to it for a certain period of time, no matter
if it is marked active and inactive.

  IMPLEMENTATION NOTE: When multiple idle destination transport
  addresses exist, it is recommended that the endpoint sends heartbeat
  messages on a Round-Robin basis, with priority given to active idle
  destination transport addresses.

The upper layer can optionally initiate the following functions:

A) disable heartbeat on a given association,
B) re-enable heart beat on a given association, and,
C) request an on-demand heartbeat on a given association.

The endpoint should keep a 'heartbeat.sent.count' counter for each
destination transport address to record the number of HEARTBEAT
messages sent to that destination transport address yet not
acknowledged upon.

When the value of this counter reaches the protocol parameter
'Max.HeartBeat.Misses', the endpoint should also mark that destination
address as inactive if it is not so marked. The endpoint may also
optionally report to the upper layer the un-reachablility of the
transport address.

The sender of the HEARTBEAT message should include in the message the
current time when the message is sent out.

The receiver of the HEARTBEAT should immediately respond with a
HEARTBEAT ACK that contains the time value copied out from the
received HEARTBEAT message.

Upon the receipt of the HEARTBEAT ACK, the sender of the HEARTBEAT
should clear the 'heartbeat.sent.count' of the destination transport
address to which the HEARTBEAT was sent, and mark the destination
transport address as active if it is not so marked. The endpoint may
also optionally report to the upper layer the reachablility of the
transport address. It also should perform an RTT measurement for that
destination transport address using the time value carried in the
HEARTBEAT ACK message.

The suggested interval for heart beat interval is 4000 ms, and may be
dynamically adjusted by adding the current RTT measurement if it is
available.

7.4 Verification Tag

Except for INIT, the sender of any SCTP datagram MUST include the
destination endpoint's Tag in the Verification Tag field of the
message. In the case of INIT, the sender should set the Verification
Tag to 0.

When sending a SHUTDOWN ACK message, the sender is allowed to either
to use the destination endpoint's Tag or fill the Verification Tag
field with 0.

When receiving an SCTP datagram (except for INIT and SHUTDOWN ACK),
the receiver MUST ensure that the value in the Verification Tag field
of the received message matches its own Tag. If the values do not
match, the receiver shall silently discard the datagram and shall NOT
process it.

The receiver of a SHUTDOWN ACK message shall accept the message
regardless the Verification Tag field is filled with the correct Tag
or 0x0.

8. Termination of Association

All existing associations should be terminated when an endpoint exits
from service. An association can be terminated by either close or
shutdown.

8.1 Close of an Association

When an endpoint decides to close down an association, it shall send
an ABORT message to its peer endpoint.

No acknowledgment is required for ABORT message. When the peer
endpoint receives the Abort, after checking the Verification Tag,
the peer shall remove the association from its record, and shall
report the termination to its upper layer.

8.2 Shutdown of an Association

An endpoint in an association may decide to gracefully shutdown the
association. This will guarantee that all outstanding datagrams from
the peer of the shutdown initiator be delivered before the association
terminates.

The initiator shall send a SHUTDOWN message to the peer of the
association, and shall include the last highest consecutive TSN it has
received from the peer in the 'Highest Consecutive TSN ACK' field. It
shall then start the T2-shutdown timer and enter the Shutdown-SENT
state. If the timer expires, the initiator must re-send the SHUTDOWN
with the updated last TSN received from its peer. When retransmitting
the SHUTDOWN, the rules in 5.3 SHALL be followed to determine the
proper timer value. The sender of the SHUTDOWN message may also
optionally include a SACK to indicate any gaps by bundling both the
SACK and SHUTDOWN message together.

Note the sender of a shutdown should limit the number of
retransmissions of the shutdown message to the protocol parameter
'Max.Retransmits'.  If Max.Retransmits is exceeded the endpoint should
destroy the TCB and may report the endpoint has unreachable to the
upper layer.

Upon the reception of the SHUTDOWN, the peer shall enter the
Shutdown-received state, and shall verify, by checking the TSN ACK
field of the message, that all its outstanding datagrams have been
received by the initiator.

If there are still outstanding datagrams left, the peer shall mark
them for retransmission and start the retransmit procedure as defined
in Section 5.3.

While in Shutdown-SENT state, the initiator shall immediately respond
to each inbound user datagram from the peer with a SACK and restart
the T2-shutdown timer.

If there is no more outstanding datagrams, the peer shall send a
SHUTDOWN ACK and then remove all record of the association.

Upon the receipt of the SHUTDOWN ACK, the initiator shall stop the
T2-shutdown timer and remove all record of the association.

Note: that it should be the responsibility of the initiator to assure
that all the outstanding datagrams on its side have been resolved
before it initiates the shutdown procedure.

Note: an endpoint shall reject any new data request from its upper
layer if it is Shutdown-SENT or Shutdown-RECEIVED state until
completion of the sequence.

Note: if an endpoint is in a Shutdown-SENT state and receives an INIT
message from its peer, it should discard the INIT message and
retransmit the shutdown message. The sender of the INIT should respond
with a stand-alone SHUTDOWN ACK in an SCTP datagram with the
Verification Tag field of its common header set to 0, and let the
normal T1-init timer cause the INIT message to be retransmitted and
thus restart the association.

9. Interface with Upper Layer

The Upper Layer Protocols (ULP) shall request for services by passing
primitives to SCTP and shall receive notifications from SCTP for
various events.

The primitives and notifications described in this section should be
used as a guideline for implementing SCTP. The following functional
description of ULP interface primitives is, at best, fictional. We
must warn readers that different SCTP implementations may have
different ULP interfaces. However, all SCTPs must provide a certain
minimum set of services to guarantee that all SCTP implementations can
support the same protocol hierarchy. This section specifies the
functional interfaces required of all SCTP implementations.

Sections 9.1 and 9.2 model interface between SCTP and the Upper Layer
Protocols. Section 9.3 models interfaces to a Layer Management entity.
The Layer Management functions could be implemented as part of the
upper layer itself or as a separate entity.

9.1 ULP-to-SCTP

The following sections functionally characterize a ULP/SCTP interface.
The notation used is similar to most procedure or function calls in
high level languages.

The ULP primitives described below specify the basic functions the
SCTP must perform to support inter-process communication. Individual
implementations must define their own exact format, and may provide
combinations or subsets of the basic functions in single calls.

A) Initialize

Format: INITIALIZE ([local port], [local eligible transport
address list])
-> local SCTP instance name

This primitive allows SCTP to initialize its internal data structures
and allocate necessary resources for setting up its operation
environment. Note that once SCTP is initialized, ULP can communicate
directly with other endpoints without re-invoking this primitive.

A local SCTP instance name will be returned to the ULP by the SCTP.

Mandatory attributes:

None.

Optional attributes:

The following types of attributes may be passed along with
the primitive:

 o local port - UDP port number, if ULP wants it to be specified;

 o local eliglible transport address list - A list of eliglible
   transport
   addresses that the local SCTP endpoint should bind. By default
   all transport interface cards should be used by the local SCTP
   host if no list is given.

B) Associate

Format: ASSOCIATE(local SCTP instance name, destination addr info,
stream count [,eligible transport address list] [,timer info])
-> association id [,destination net list] [,outbound stream count]

This primitive allows the upper layer to initiate an association to a
specific peer endpoint. The peer endpoint shall be specified by one of
the transport addresses which define the endpoint (see section 1.1).
If the local SCTP instance has not been initialized, the ASSOCIATE is
considered an error. The set of transport addresses specified in the
"eligible transport address list" shall be used has valid destinations
when sending to the peer endpoint. If this parameter is not specified,
the associate command will consider all of the transport addresses
returned by the INIT ACK message has valid.

An association id, which is a local handle to the SCTP association,
will be returned on successful establishment of the association.  If
SCTP is not able to open an SCTP association with the peer endpoint,
an error is returned.

Implementor's Note: If ASSOCIATE primitive is implemented as a
blocking function call, the ASSOCIATE primitive can return association
parameters in addition to the association id upon successful
establishment. If ASSOCIATE primitive is implemented as a non-blocking
call, only the association id shall be returned and association
parameters shall be passed using the COMMUNICATION UP notification.

The association parameters shall include the destination addresses of
the peer as well as the outbound stream count.  One of the transport
address from the set of destination addresses will be used as default
primary destination address for sending datagrams to this peer. The
returned "destination net list" can be used by the ULP to override the
default primary destination transport address or to force sending a
datagram on a specific network.

Mandatory attributes:

 o local SCTP instance name - obtained from the initialize operation.

 o destination addr info - specified as one of the transport addresses
   of the peer endpoint with which the association is to be established.

 o stream count - the number of streams the ULP would like to open at the
   beginning of the association.

Optional attributes:

 o eligible transport address list - a list of transport addresses
   that the endpoint is allowed to use for sending datagrams to the
   peer. By default, all transport addresses of the peer are
   available.

 o timer info - Timer selection and its operation syntax -- to indicate
   to SCTP an alternative timer the SCTP should use for its operation.

C) Terminate

Format: TERMINATE(association id)
-> result

Gracefully terminates an association. Any locally queued datagrams
will be delivered to the peer. The association will be terminated only
after the peer acknowledges all the messages sent.  A success code
will be returned on successful termination of the association. If
attempting to terminate the association results in a failure, an error
code shall be returned.

Mandatory attributes:

 o association id - local handle to the SCTP association

Optional attributes:

None.

D) Abort

Format: ABORT(association id)
-> result

Ungracefully terminates an association. Any locally queued datagrams
will be discarded and an ABORT message is sent to the peer.  A success
code will be returned on successful abortion of the association. If
attempting to abort the association results in a failure, an error
code shall be returned.

Mandatory attributes:

 o association id - local handle to the SCTP association

Optional attributes:

None.

E) Send

Format: SEND(association id, buffer address, byte count
[,context] [,stream id] [,life time] [,destination transport address]
[,un-order flag] [,no-bundle flag] )

This is the main method to send datagrams via SCTP.

Mandatory attributes:

 o association id - local handle to the SCTP association

 o buffer address - the location where the payload to be transmitted is
   stored;

 o byte count - The size of the payload in number of octets;

Optional attributes:

 o context - optional information that will be carried in the
   sending failure notification to the ULP if the transportation of
   this datagram fails.

 o stream id - to indicate which stream to send the data on. If not
   specified, stream 0 will be used.

 o life time - specifies the life time of the message. The message
   will not be sent by SCTP after the life time expires. This
   parameter can be used to avoid efforts to transmit stale
   datagrams. SCTP notifies the ULP, if the datagram cannot be
   initiated to transport (i.e. sent to the destination via SCTP's
   send primitive) within the life time variable. How ever, the
   message will be transmitted if a TSN has been assigned to the
   message before the life time expired.

 o destination transport address - specified as one of the destination
   transport addresses of the peer endpoint to which this message
   should be sent. Whenever possible, SCTP should use this destination
   transport address for sending the datagram, instead of the current
   primary destination transport address.

 o un-order flag - this flag, if present, indicates that the user
   would like the data delivered in an un-ordered fashion to the
   remote peer.

 o no-bundle flag - Instructs SCTP not to bundle the user data with
   other outbound DATA chunks. Note: SCTP may still bundle even when
   this flag is present, when faced with network congestion.

F) Set Primary

Format: SETPRIMARY(association id, destination transport address)
-> result

Instructs the local SCTP to use the specified destination transport
address as primary destination address for sending datagrams.

The result of attempting this operation shall be returned. If the
specified destination transport address is not present in the
"destination transport address list" returned earlier in an associate
command or communication up notification, an error shall be returned.

Mandatory attributes:

 o association id - local handle to the SCTP association

 o destination transport address - specified as one of the transport
   addresses of the peer endpoint, which should be used as primary
   address for sending datagrams. This overrides the current primary
   address information maintained by the local SCTP endpoint.

G) Receive

Format: RECEIVE(association id, buffer address, buffer size [,stream id])
-> byte count [,transport address] [,stream id] [,sequence number]

This primitive shall read the first datagram in the SCTP in-queue to
ULP, if there is one available, into the specified buffer. The size of
the datagram read, in octets, will be returned. It may, depending on
the specific implementation, also return other information such as the
sender's address, the stream id on which it is received, whether there
are more datagrams available for retrieval, etc. For ordered messages,
their sequence number may also be returned.

Depending upon the implementation, if this primitive is invoked when
no datagram is available the implementation should return an
indication of this condition or should block the invoking process
until data does become available.

Mandatory attributes:

 o association id - local handle to the SCTP association

 o buffer address - the memory location indicated by the ULP to store
 the received datagram.

 o buffer size - the maximum size of data to be received, in octets.

Optional attributes:

 o stream id - to indicate which stream to receive the data on.

H) Status

Format: STATUS(association id) -> status data

This primitive shall return a data block containing the following
information:
  receive window size,
  send window size,
  connection state,
  number of buffers awaiting acknowledgement,
  number of buffers pending receipt,
  primary destination address,
  round trip time on primary destination address,
  retransmission time out value on primary destination address,
  other destination addresses,
  round trip times on other destination addresses.

Mandatory attributes:

 o association id - local handle to the SCTP association

Optional attributes:

 None.

I) Change Heartbeat

Format: CHANGEHEARTBEAT(association id, new state)
-> result

Instructs the local SCTP to enable or disable heart beat on the
specified association.

The result of attempting this operation shall be returned.

Mandatory attributes:

 o association id - local handle to the SCTP association

 o new state - the new state of heart beat for this association (either
   enabled or disabled).

J) Request HeartBeat

Format: REQUESTHEARTBEAT(association id, transport address)

Instructs the local SCTP to perform a HeartBeat on the specified
transport address of the given association. The results of the
HeartBeat should update the RTT information.

Mandatory attributes:

 o association id - local handle to the SCTP association

 o transport address - the transport address of the association
   on which a heartbeat should be issued.

K) Get RTT Report

Format: GETRTTREPORT(association id, transport address)
-> rtt result

Instructs the local SCTP to report the current RTT measurement on the
specified transport address of the given association. The returned
result can be an intager containing the most recent RTT in
milliseconds.

Mandatory attributes:

 o association id - local handle to the SCTP association

 o transport address - the transport address of the association
   on which the RTT measurement is to be reported.

9.2 SCTP-to-ULP

It is assumed that the operating system or application environment
provides a means for the SCTP to asynchronously signal the ULP
process. When SCTP does signal an ULP process, certain information is
passed to the ULP.

A) DATA ARRIVE notification

SCTP shall invoke this notification on the ULP when a datagram is
successfully received and ready for retrieval.

The following may be optionally be passed with the notification:

 o association id - local handle to the SCTP association

 o stream id - to indicate which stream the data is received on.

B) SEND FAILURE notification

If a datagram can not be delivered SCTP shall invoke this notification
on the ULP.

The following may be optionally be passed with the notification:

 o data - the location ULP can find the un-delivered datagram.

 o context - optional information associated with this datagram (see
   D in section 9.1).

 o association id - local handle to the SCTP association

C) NETWORK STATUS CHANGE notification

When a destination transport address is marked down (e.g., when SCTP
detects a failure), or marked up (e.g., when SCTP detects a recovery),
SCTP shall invoke this notification on the ULP.

The following shall be passed with the notification:

 o destination transport address - This indicates the destination
   transport address of the peer endpoint affected by the change;

 o new-status - This indicates the new status.

D) COMMUNICATION UP notification

This notification is used when SCTP becomes ready to send or receive
datagrams, or when a lost communication to an endpoint is restored.

  IMPLEMENTATION NOTE: If ASSOCIATE primitive is implemented as a
  blocking function call, the association parameters are returned as a
  result of the ASSOCIATE primitive itself. In that case,
  COMMUNICATION UP notification is optional at the association
  initiator's side.

The following shall be passed with the notification:

 o status - This indicates what type of event that has occurred;

 o association id - local handle to the SCTP association

 o destination transport address list - the set of transport addresses
   of the peer

 o outbound stream count - the maximum number of streams allowed to be
   used in this association by the ULP

E) COMMUNICATION LOST notification

When SCTP loses communication to an endpoint completely or detects
that the endpoint has performed an abort or graceful shutdown
operation, it shall invoke this notification on the ULP.

The following shall be passed with the notification:

 o status - This indicates what type of event that has occurred;

 o association id - local handle to the SCTP association

The following may be optionally passed with the notification:

 o unsent-datagrams - The number and location of un-sent datagrams
   still in hold by SCTP;

 o unacknowledged-datagrams - The number and location of datagrams
   that were attempted to be transported to the destination, but were
   not acknowledged when the loss of communication was detected.

 o last-acked - the sequence number last acked by that peer endpoint;

 o last-sent - the sequence number last sent to that peer endpoint;

 o received-but-not-delivered - datagrams that were received by SCTP
   but not yet delivered to the ULP.

Note: the un-send data report may not be accurate for those user
messages which are segmented by SCTP during transmission.

9.3 Interfaces to Layer Management

This section models interfaces to a Layer Management (LM) entity,
which manages resources that have transport layer-wide impact. Layer
Management consists of primitives related to the management of SCTP
performed as a subset of systems management.

9.3.1 LM-to-SCTP

The following ULP-to-SCTP primitives from section 9.1 could be
implemented as part of the LM entity.

INITIALIZE
SETPRIMARY
ABORT
STATUS

9.3.2 SCTP-to-LM

The following SCTP-to-ULP primitives from section 9.2 could be implemented
as part of the LM entity.

SEND FAILURE notification
NETWORK STATUS CHANGE notification
COMMUNICATION UP notification
COMMUNICATION LOST notification

10. Security Considerations

10.1 Security Objectives

As a common transport protocol designed to reliably carry time-
sensitive user messages, such as billing or signalling messages for
telephony services, between two networked endpoints, SCTP has the
following security objectives.
  - availability of reliable and timely data transport services
  - integrity of the user-to-user information carried by SCTP

10.2 SCTP Responses To Potential Threats

It is clear that SCTP may potentially be used in a wide variety of
risk situations.  It is important for operator(s) of the SCTP Hosts
concerned to analyze their particular situations and decide on the
appropriate counter-measures.

Where the SCTP Host serves a group of users, it is probably
operating as part of a professionally managed corporate or service
provider network.  It is reasonable to expect that this management
includes an appropriate security policy framework.  [RFC 2196, "Site
Security Handbook", B. Fraser Ed., September 1997] should be
consulted for guidance.

The case is more difficult where the SCTP Host is operated by a
private user.  The service provider with whom that user has a
contractual arrangement SHOULD provide help to ensure that the
user's site is secure, ranging from advice on configuration through
downloaded scripts and security software.

10.2.1 Countering Insider Attacks

The principles of the Site Security Handbook [ ] should be applied
to minimize the risk of theft of information or sabotage by
insiders.  These include publication of security policies, control
of access at the physical, software, and network levels, and
separation of services.

10.2.2 Protecting against Data Corruption in the Network

Where the risk of undetected errors in datagrams delivered by the
lower layer transport services is considered to be too great,
additional checksum protection may be required.  The question is
whether this is appropriately provided as an SCTP service because it
is needed by most potential users of SCTP, or whether instead it
should be provided by the SCTP user application.  (The SCTP protocol
overhead, as opposed to the signalling payload, is protected
adequately by the UDP checksum and measures taken in SCTP to prevent
replay attacks and masquerade.)  In any event, the checksum must be
specifically designed to ensure that it detects the errors left
behind by the UDP checksum.

10.2.3 Protecting Confidentiality

In most cases, the risk of breach of confidentiality applies to the
signalling data payload, not to the SCTP or lower-layer protocol
overheads.  If that is true, encryption  of the SCTP user data only
may be considered.  As with the supplementary checksum service, user
data encryption may be performed either by the SCTP user application
or as a service of SCTP itself.  If it is performed by SCTP, the
user data must be encrypted before any checksum is applied.

Particularly for mobile users, the requirement for confidentiality
may include the masking of IP addresses and ports.  In this case
IPSEC ESP should be used instead of application-level encryption.
Similarly, where other reasons prompt the use of the IPSEC ESP
service, application-level encryption is unnecessary. It will be up
to the SCTP Host operators to configure the application
appropriately.

Regardless of which level performs the encryption, the IPSEC ISAKMP
service should be used for key management.

Operators should consult [RFC 2401, "Security Architecture for the
Internet Protocol", S. Kent, R. Atkinson, November 1998] for
information on the configuration of IPSEC services between hosts
with and without intervening firewalls.

10.2.4 Protecting against Blind Denial of Service Attacks

A blind attack is one where the attacker is unable to intercept or
otherwise see the content of data flows passing to and from the
target SCTP Host where it is not a party to the association.  Blind
denial of service attacks may take the form of flooding, masquerade,
or improper monopolization of services.

10.2.4.1 Flooding

The objective of flooding is to cause loss of service and incorrect
behaviour at target systems through resource exhaustion, interference
with legitimate transactions, and exploitation of buffer-related
software bugs.  Flooding may be directed either at the SCTP Host or at
resources in the intervening IP Access Links or the Internetwork.
Where the latter entities are the target, flooding will manifest
itself as loss of network services, including potentially the breach
of any firewalls in place.

In general, protection against flooding begins at the equipment
design level, where it includes measures such as:
 - avoiding commitment of limited resources before determining that
   the request for service is legitimate
 - giving priority to completion of processing in progress over the
   acceptance of new work
 - identification and removal of duplicate or stale queued requests
   for service.

Network equipment should be capable of generating an alarm and log
if a suspicious increase in traffic occurs.  The log should provide
information such as the identity of the incoming link and source
address(es) used which will help the network or SCTP Host operator
to take protective measures.  Procedures should be in place for the
operator to act on such alarms if a clear pattern of abuse emerges.

The design of SCTP is resistant to flooding attacks, particularly in
its use of a four-way start-up handshake, its use of a cookie to
defer commitment of resources at the responding SCTP Host until the
handshake is completed, and its use of a verification tag to prevent
insertion of extraneous messages into the flow of an established
association.

10.2.4.2 Masquerade

Masquerade can be used to deny service in several ways:
 - by tying up resources at the target SCTP Host to which the
   impersonated host has limited access.  For example, the target host
   may by policy permit a maximum of one SCTP association with the
   impersonated SCTP Host.  The masquerading attacker may attempt to
   establish an association purporting to come from the impersonated
   host so that the latter cannot do so when it requires it.
 - by deliberately allowing the impersonation to be detected,
   thereby provoking counter-measures which cause the impersonated host
   to be locked out of the target SCTP Host
 - by interfering with an established association by inserting
   extraneous content such as a SHUTDOWN request.

SCTP prevents masquerade through IP spoofing by use of the four-way
startup handshake.  Because the initial exchange is memoryless, no
lockout mechanism is triggered by masquerade attacks.  SCTP protects
against insertion of extraneous messages into the flow of an
established association by use of the verification tag.

Logging of received INIT requests and abnormalities such as
unexpected INIT ACKs might be considered as a way to detect patterns
of hostile activity.  However, the potential usefulness of such
logging must be weighed against the increased SCTP startup
processing it implies, rendering the SCTP Host more vulnerable to
flooding attacks.  Logging is pointless without the establishment of
operating procedures to review and analyze the logs on a routine
basis.

10.2.4.3 Improper Monopolization of Services

Attacks under this heading are performed openly and legitimately by
the attacker.  They are directed against fellow users of the target
SCTP Host or of the shared resources between the attacker and the
target host.  Possible attacks include the opening of a large number
of associations between the attacker's host and the target, or
transfer of large volumes of information within a legitimately-
established association.

Such attacks take advantage of policy deficiencies at the target
SCTP Host.  Defense begins with a contractual prohibition of
behaviour directed to denial of service to others.  Policy limits
should be placed on the number of associations per adjoining SCTP
Host.  SCTP user applications should be capable of detecting large
volumes of illegitimate or "no-op" messages within a given
association and either logging or terminating the association as a
result, based on local policy.

10.3 Protection against Fraud and Repudiation

The objective of fraud is to obtain services without authorization
and specifically without paying for them.  In order to achieve this
objective, the attacker must induce the SCTP user application at the
target SCTP Host to provide the desired service while accepting
invalid billing data or failing to collect it.  Repudiation is a
related problem, since it may occur as a deliberate act of fraud or
simply because the repudiating party kept inadequate records of
service received.

Potential fraudulent attacks include interception and misuse of
authorizing information such as credit card numbers, blind
masquerade and replay, and man-in-the middle attacks which modify
the messages passing through a target SCTP association in real time.

The interception attack is countered by the confidentiality measures
discussed in section 10.2.3 above.

Section 10.2.4.2 describes how SCTP is resistant to blind masquerade
attacks, as a result of the four-way startup handshake and the
validation tag.  The validation tag and TSN together are protections
against blind replay attacks, where the replay is into an existing
association.

However, SCTP does not protect against man-in-the-middle attacks
where the attacker is able to intercept and alter the messages sent
and received in an association.  Where a significant possibility of
such attacks is seen to exist, or where possible repudiation is an
issue, the use of the IPSEC AH service is recommended to ensure both
the integrity and the authenticity of the messages passed.

SCTP also provides no protection against attacks originating at or
beyond the SCTP Host and taking place within the context of an
existing association.  Prevention of such attacks should be covered
by appropriate security policies at the host site, as discussed in
section 10.2.1.

11. IANA Consideration

This protocol may be extended through IANA in three ways:
 -- through definition of additional chunk types,
 -- through definition of additional parameter types, or
 -- through definition of additional cause codes within Operation
    Error chunks

11.1 IETF-defined Chunk Extension

The appropriate use of specific chunk types is an integral part of the
SCTP protocol.  In consequence, the intention is that new IETF-defined
chunk types MUST be supported by standards-track RFC documentation.
As a transitional step, a new chunk type MAY be introduced in an
Experimental RFC. Chunk type codes MUST remain permanently associated
with the original documentation on the basis of which they were
allocated.  Thus if the RFC supporting a given chunk type is
deprecated in favour of a new document, the corresponding chunk type
code value is also deprecated and a new code value is allocated in
association with the replacement document.

The documentation for a new chunk code type must include the following
information:
(a) a long and short name for the new chunk type;
(b) a detailed description of the structure of the chunk, which MUST
    conform to the basic structure defined in section 2.2;
(c) a detailed definition and description of intended use of each field
    within the chunk, including the chunk flags if any;
(d) a detailed procedural description of the use of the new chunk type
    within the operation of the protocol.

If the primary numbering space reserved for IETF use (0x00 to 0xFD) is
exhausted, new codes shall subsequently be allocated in the extension
range 0x0000 through 0xFFFF. Chunks allocated in this range MUST
conform to the following structure:

First word (32 bits):
  as shown in section 2.2, with chunk type code equal to 0xFF.

Second word:
  first octet MUST be all 1's (0xFF). Next octet MUST be all 0's
  (0x00).  Final two octets contain the allocated extension code value.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1 1 1 1 1 1 1 1|Chunk  Flags   |      Chunk Length             |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1 1 1 1 1 1 1 1|0 0 0 0 0 0 0 0|    Extension Type Code        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   \                                                               \
   /                    Value                                      /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

11.2 IETF-defined Chunk Parameter Extension

The allocation of a new chunk parameter type code from the IETF
numbering space MUST be supported by RFC documentation.  As with chunk
type codes, parameter type codes are uniquely associated with their
supporting document and MUST be replaced if new documentation is
provided.  This documentation may be Informational, Experimental, or
standards-track at the discretion of the IESG.  It MUST contain the
following information:
(a) Name of the parameter type.
(b) Detailed description of the structure of the parameter field. This
    structure MUST conform to the general type-length-value format
    described in section 2.2.1.
(c) Detailed definition of each component of the parameter value.
(d) Detailed description of the intended use of this parameter type,
    and an indication of whether and under what circumstances
    multiple instances of this parameter type may be found within the
    same chunk.

Additional parameter type codes may be allocated initially from the
range 0x0000 through 0xFFFD.  If this space is exhausted, extension
codes shall be allocated in the range 0x0000 through 0xFFFF.  Where an
extension code has been allocated, the format of the parameter must
conform to the following structure:

First word (32 bits):
  contains the parameter type code 0xFFFF and parameter length as
  described in section 2.2.1.

Second word:
  first octet MUST be all 1's (0xFF).  Next octet MUST be all 0's
  (0x00). Final two octets contain the allocated extension code
  value.

The Value portion of the parameter, if any, follows the second word.

    0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1|             Length            |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |1 1 1 1 1 1 1 1|0 0 0 0 0 0 0 0|    Extension Type Code        |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   \                                                               \
   /                    Value                                      /
   \                                                               \
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

11.3 IETF-defined Additional Error Causes

Additional cause codes may be allocated in the range 0x0004 to 0xFFFF
upon receipt of any permanently-available public documentation
containing the following information:
(a) Name of the error condition.
(b) Detailed description of the conditions under which an SCTP
    endpoint should issue an Operation Error with this cause code.
(c) Expected action by the SCTP endpoint which receives an Operation
    Error chunk containing this cause code.
(d) Detailed description of the structure and content of data fields
    which accompany this cause code.

The initial word (32 bits) of a cause code parameter MUST conform to
the format shown in section 2.3.9, i.e.:
 -- first two octets contain the cause code value
 -- last two octets contain length of the cause parameter.

12. Suggested SCTP Protocol Parameter Values

The following protocol parameters are recommended:

RTO.Initial              -  3 seconds
Valid.Cookie.Life        -  5 seconds
Max.Retransmits          - 10 attempts
Max.Init.Retransmit      -  8 attempts
Max.HeartBeat.Misses     -  3 attempts

Miscellaneous protocol variables/counters:

'retrans.count'        - per association counter
'heartbeat.sent.count' - per destination transport address counter

13. Acknowledgments

The authors wish to thank Mark Allman, Richard Band, Scott Bradner,
Ram Dantu, R. Ezhirpavai, Sally Floyd, Matt Holdrege, Henry Houh, Gary
Lehecka, Lyndon Ong, Kelvin Porter, Heinz Prantner, Jarno Rajahalme,
A. Sankar, Greg Sidebottom, Brian Wyld, and many others for their
invaluable comments.

14.  Authors' Addresses

Randall R. Stewart                      Tel: +1-847-632-7438
Motorola, Inc.                          EMail: rstewar1@email.mot.com
1501 W. Shure Drive, #2315
Arlington Heights, IL 60004
USA

Qiaobing Xie                            Tel: +1-847-632-3028
Motorola, Inc.                          EMail: qxie1@email.mot.com
1501 W. Shure Drive, #2309
Arlington Heights, IL 60004
USA

Ken Morneault                           Tel: +1-703-484-3323
Cisco Systems Inc.                      EMail:kmorneau@cisco.com
13615 Dulles Technology Drive
Herndon, VA. 20171
USA

Chip Sharp                              Tel: +1-919-472-3121
Cisco Systems Inc.                      EMail:chsharp@cisco.com
7025 Kit Creek Road
Research Triangle Park, NC  27709
USA

Hanns Juergen Schwarzbauer              Tel: +49-89-722-24236
SIEMENS AG
Hofmannstr. 51
81359 Munich
Germany
EMail: HannsJuergen.Schwarzbauer@icn.siemens.de

Tom Taylor                              Tel: +1-613-736-0961
Nortel Networks
1852 Lorraine Ave.
Ottawa, Ontario
Canada K1H 6Z8
EMail:taylor@nortelnetworks.com

Ian Rytina                              Tel:
Ericsson Australia                      EMail:ian.rytina@ericsson.com
37/360 Elizabeth Street
Melbourne, Victoria 3000
Australia

Malleswar Kalla                         Tel: +1-973-829-5212
Telcordia Technologies
MCC 1J211R
445 South Street
Morristown, NJ 07960
USA
EMail: kalla@research.telcordia.com

Lixia Zhang                             Tel: +1-310-825-2695
UCLA Computer Science Department        EMail: lixia@cs.ucla.edu
4531G Boelter Hall
Los Angeles, CA 90095-1596
USA

Vern Paxson                             Tel: +1-510-642-4274 x 302
ACIRI                                   EMail: vern@aciri.org
1947 Center St., Suite 600,
Berkeley, CA 94704-1198
USA

15. References

[1] Eastlake , D. (ed.), "Randomness Recommendations for Security",
    RFC 1750, December 1994.

[2] ITU-T Recommendation Q.703 "Q.703 - Signaling link", July 1996.

[3] Allman, M., Paxson, V., and Stevens, W., "TCP Congestion
    Control", RFC 2581, April 1999.

[4] Rivest, R., "The MD5 Message-Digest Algorithm", RFC 1321,
    August 1999.

[5] M. Allman and V. Paxson, "On Estimating End-to-End Network Path
    Properties", Proc. SIGCOMM '99, 1999.

      This Internet Draft expires in 6 months from October 1999.