Benchmarking Working Group              H.Berkowitz, Gett Communications
Internet Draft                                          S.Hares, Next Hop Nexthop
Document: draft-ietf-bmwg-conterm-01.txt draft-ietf-bmwg-conterm-02.txt                 A.Retana, Cisco
Expires August December 2002                         P.Krishnaswamy, Consultant
                                                M.Lepp, Juniper Networks
                                               E.Davies, Nortel Networks
June 2002

          Terminology for Benchmarking
               External Routing BGP Device Convergence Measurements
                          in the Control Plane

Status of this Memo

   This document is an Internet-Draft and is in full conformance with
   all provisions of Section 10 of RFC2026 [1].
   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.
   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time. It is inappropriate to use Internet- Drafts as reference
   material or to cite them other than as "work in progress."

   The list of current Internet-Drafts can be accessed at
   The list of Internet-Draft Shadow Directories can be accessed at

   A revised version of this draft document will be submitted to the RFC
   editor as a Informational document for the Internet Community.
   Discussion and suggestions for improvement are requested.
   This document will expire before August December 2002. Distribution of this
   draft is unlimited.

Copyright Notice

   Copyright (C) The Internet Society (2002).  All Rights Reserved.


   This draft establishes terminology to standardize the description of
   benchmarking methodology for measuring eBGP convergence in the
   control plane of a single BGP device. Future documents will address
   iBGP convergence, the initiation of forwarding based on converged
   control plane information and internet-wide convergence.

Conventions used in this document

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED",  "MAY", multiple interacting BGP devices. This
   terminology is applicable to both IPv4 and "OPTIONAL" in this
   document IPv6. Illustrative
   examples of each version are to be interpreted as described in RFC-2119 [2].

 Berkowitz, et al                                                     1 included where relevant.

   Table of Contents
   1. Introduction....................................................3
      1.1  Overview and Roadmap.......................................3
      1.2  Definition Format..........................................3
   2. Constituent elements of a router or network  of routers.........4
      2.1  BGP Peer...................................................4 Instance or Device.....................................4
      2.2  BGP Peer...................................................5
      2.3  Default Route, Default Free Table, and Full Table..........5
      2.4  Classes of BGP-Speaking Routers............................7 Routers............................8
   3. Routing Data Structures.........................................9
      3.1  Routing Information Base (RIB).............................9
      3.2  Policy....................................................10  Policy....................................................11
      3.3  Policy Information Base...................................11
      3.4  The  Forwarding Information Base (FIB).....................12 (FIB).........................12
   4. Components and characteristics of Routing information..........12 information..........13
      4.1  Prefix....................................................12  (Network) Prefix..........................................13
      4.2  Route.....................................................13  Network Prefix Length.....................................13
      4.3  BGP Route.................................................13  Route.....................................................14
      4.4  Route Instance............................................14  BGP Route.................................................14
      4.5  Active Route..............................................14  BGP Route Attributes and BGP Timers.......................15
      4.6  Unique Route..............................................14  Route Instance............................................16
      4.7  Non-Unique Route..........................................15  Active Route..............................................17
      4.8  Route Packing.............................................15  Unique Route..............................................17
      4.9  Route Mixture.............................................15  Non-Unique Route..........................................17
      4.10    Update Train...........................................16    BGP UPDATE message.....................................17
      4.11    Characterization of sets of update messages............18
      4.12    Route Flap.............................................18 Flap.............................................20
   5. Route Changes and Convergence..................................18 Convergence..................................21
      5.1  Route Change Events.......................................18 Events.......................................21
      5.2  Convergence...............................................19  Device Convergence in the Control Plane...................22
   6. BGP Operation Events...........................................20 Events...........................................23
      6.1  Hard reset................................................20 reset................................................24
      6.2  Soft reset................................................21 reset................................................24
   7. Factors that impact the performance of the convergence process.21 process.24
      7.1  General factors affecting BGP convergence.................21 device convergence..............24
      7.2  Implementation-specific and other factors affecting BGP
   8. Security Considerations........................................23 Considerations........................................27
   9. References.....................................................24 References.....................................................27
   10. Acknowledgments...............................................25 Acknowledgments...............................................28
   11. Author's Addresses............................................25

Berkowitz, et al            Expires: August 2002                      2 Addresses............................................29

1. Introduction

   This document defines terminology for use in characterizing the
   convergence performance of BGP processes in routers or other devices
   that instantiate BGP functionality. functionality (see RFC1771 [1]). It is the first
   part of a two document series, of which the subsequent document will
   contain the associated tests and methodology. This terminology is
   applicable to both IPv4 and IPv6. Illustrative examples of each
   version are included where relevant.

   The following observations underlie the approach adopted in this, and
   the companion document:
   -  The principal objective is to derive methodologies to standardize
      conducting and reporting convergence-related measurements for BGP.
   -  It is necessary to remove ambiguity from many frequently used
      terms that arise in the context of such measurements.
   -  As convergence characterization is a complex process, it is
      desirable to restrict the initial focus in this set of documents
      to specifying how to take basic control plane measurements as a
      first step to characterizing BGP convergence.

   For path vector protocols protocols, such as BGP, the primary initial focus
   will therefore be on network and system control-plane activity
   consisting of the arrival, processing, and propagation of routing
   Subsequent drafts will explore the more intricate aspects of
   convergence measurement, such as the impacts of the presence of
   policy processing, simultaneous traffic on the control and data paths
   within the DUT, and other realistic performance modifiers.
   Convergence of Interior Gateway Protocols will also be considered in
   separate drafts.

1.1 Overview and Roadmap

   Characterizations of the BGP convergence performance of a device must
   take into account all distinct stages and aspects of BGP
   functionality. This requires that the relevant terms and metrics be
   as specifically defined as possible. Such definition is the goal of
   this document.

   The necessary definitions are classified into two separate
   -  Descriptions of the constituent elements of a network or a router
      that is undergoing convergence
   -  Descriptions of factors that impact convergence processes

1.2 Definition Format

   The definition format is equivalent to that defined in [5], [3], and is
   repeated here for convenience:

   X.x Term to be defined. (e.g., Latency)

              The specific definition for

              One or more sentences forming the term being defined.

Berkowitz, et al            Expires: August 2002                      3 body of the definition.

              A brief discussion of the term, its application and any
              restrictions that there might be on measurement

   Measurement units:
              The units used to report measurements of this term, if
              List of issues or conditions that could affect this term.

   See Also:
              List of related terms that are relevant to the definition
              or discussion of this term.

2. Constituent elements of a router or network  of routers.

   Many terms included in this list of definitions were originally
   described in previous standards or papers. They are included here
   because of their pertinence to this discussion. Where relevant,
   reference is made to these sources. An effort has been made to keep
   this list complete with regard to the necessary concepts without over

2.1 BGP Instance or Device

              A BGP instance is a process with a single Loc-RIB that
              runs on a BGP device.

              We have chosen to use "device" as the general case, to
              deal with the understood [e.g. [9]] and yet-to-be-invented
              cases where the control processing may be separate from
              forwarding [12].  A BGP device may be a traditional
              router, a route server, a BGP-aware traffic steering
              device, a device using BGP to exchange topology
              information with a GMPLS environment, etc. A device such
              as a route server, for example, never forwards traffic, so
              forwarding-based measurements would be meaningless for it.

   Measurement units: N/A


   See Also:

2.2 BGP Peer

              A BGP peer is another BGP instance to which the Device
              Under Test (DUT) has established a TCP connection over
              which a BGP session is active.  In the test scenarios in
              the methodology discussion that will follow this draft,
              peers send BGP advertisements to the DUT and receive DUT-
              originated advertisements.

              This is a protocol-specific definition, not to be confused
              with another frequent usage, which refers to the
              business/economic definition for the exchange of routes
              without financial compensation.

              It is worth noting that a BGP peer, by this definition is
              associated with a BGP peering session, and there may be
              more than one such active session on a router or on a
              tester.  The peering sessions referred to here may exist
              between various classes of BGP routers (see section 2.3).

   Measurement units: number of BGP peers


   See Also:

Berkowitz, et al            Expires: August 2002                      4


2.3 Default Route, Default Free Table, and Full Table

   An individual router's routing table may not necessarily contain a
   default route.  Not having a default route, however, is not
   synonymous with having a full default-free table(DFT).
   It should be noted that the references to number of routes in this
   section are to routes installed in the loc-RIB, not route instances,
   and that the total number of route instances may be 4 to 10 times the
   number of routes.

   The actual path setup and forwarding of MPLS speaking routers are
   outside the scope of this document.  A device that computes BGP
   routes that may give
   routes, which a sub-IP device information that it uses can use to set up paths, however, has its BGP
   aspects within scope.


2.3.1 Default Route

              A Default Route is a route entry that can match any
              prefix. If a router does not have a route for a particular
              packet's destination address, it forwards this packet to
              the next hop in the default route entry, provided its
              Forwarding Table (Forwarding Information Base (FIB)
              contains one. The notation for a default route for IPv4 is
     and for IPv6 it is 0:0:0:0:0:0:0:0 or ::/0.


   Measurement units: N.A.


   See Also: default free routing table, route, route instance

Berkowitz, et al            Expires: August 2002                      5


2.3.2 Default Free Routing Table

              A default free routing table with has no default routes, as routes and is
              typically seen in routers in the core or top tier of
              routers in the network.

              The term originates from the concept that routers at the
              core or top tier of the Internet will not be configured
              with a default route (Notation in IPv4 and in
              IPv6 0:0:0:0:0:0:0:0 or ::/0). Thus they will forward
              every prefix to a specific next hop based on the longest
              match on the IP addresses.

              Default free routing table size is commonly used as an
              indicator of the magnitude of reachable Internet address
              space. However, default free routing tables may also
              include routes internal to the router's AS.

   Measurements: The number of routes

   See Also: Full Default Free, Default Route


2.3.3 Full Default Free Table

              A set of BGP full default free table is a set of BGP routes generally
              accepted to be the complete set of BGP routes collectively
              announced by the complete set of autonomous systems making
              up the public Internet.  Due to the dynamic nature of the
              Internet, the exact size and composition of this table may
              vary slightly depending where and when it is observed.
              Several investigators ([12],[13][14]) ([17],[18],[19]) measure this on a
              daily and/or weekly basis; June 2001 measurements put the
              table at approximately 105,000 routes, growing

              It is generally accepted that a full table, in this usage,
              does not contain the infrastructure routes or individual
              sub-aggregates of routes that are otherwise aggregated by
              the provider before announcement to other autonomous

   Measurement Units: number of routes


   See Also: Routes, Route Instances, Default Route

Berkowitz, et al            Expires: August 2002                      6


2.3.4 Full Provider Internal Table

              A full provider internal table is a superset of the full
              routing table that contains infrastructure and non-aggregated non-
              aggregated routes.

              Experience has shown that this table can contain 1.3 to
              1.5 times the number of routes in the externally visible
              full table.  Tables of this size, therefore, are a real-
              world requirement for key internal provider routers.

   Measurement Units: number of routes


   See Also: Routes, Route Instances, Default Route


2.4 Classes of BGP-Speaking Routers

   A given router may perform more than one of the following functions,
   based on its logical location in the network.


2.4.1 Provider Edge Router

              A provider edge router is a router at the edge of a
              provider's network, configured to speak BGP, which peers
              with a BGP speaking router operated by the end-user. The
              traffic that transits this router may be destined to, or
              originate from non-contiguous autonomous systems.

              Such a router will always speak eBGP and may speak iBGP.

   Measurement units:


   See Also:

Berkowitz, et al            Expires: August 2002                      7


2.4.2 Subscriber Edge Router

              A subscriber edge router is a BGP-speaking router
              belonging to an end user organization that may be multi-homed, multi-
              homed, and which carries traffic only to and from that end
              user AS.

              Such a router will always speak eBGP and may speak iBGP.

   Measurement units:


   See Also:

2.3.3 Interprovider

2.4.3 Inter-provider Border Router

              An inter-provider border router is a BGP speaking router
              which maintains BGP sessions with another BGP speaking
              router in another provider AS.  Traffic transiting this
              router may be directed to or from another AS that has no
              direct connectivity with this provider's AS.

              Such a router will always speak eBGP and may speak iBGP.

   Measurement units:


   See Also:

2.3.4 Intraprovider

2.4.4 Intra-provider Core Router

              An intra-provider core router is a provider router
              speaking iBGP to the provider's edge routers, other intraprovider intra-
              provider core routers, or the provider's interprovider inter-provider
              border routers.

              Such a router will always speak iBGP and may speak eBGP.

   Measurement units:

              MPLS speaking routers are outside the scope of this
              document.  It is entirely likely, however, that core Label
              Switched Routers, especially in the P router role of
              RFC 2547 [16], [10], may contain little or no BGP information.

   See Also:

Berkowitz, et al            Expires: August 2002                      8

3. Routing Data Structures

3.1 Routing Information Base (RIB)

   The RIB collectively consists of a set of logically (not necessarily
   literally) distinct databases, each of which is enumerated below. The
   RIB contains all destination prefixes to which the router may
   forward, and one or more currently reachable next hop addresses for

   Routes included in this set potentially have been selected from
   several sources of information, including hardware status, interior
   routing protocols, and exterior routing protocols. RFC 1812 contains
   a basic set of route selection criteria relevant in an all-source
   context. Many implementations impose additional criteria.  A common
   implementation-specific criterion is the preference given to
   different routing information sources.

3.1.1 Adj-RIB-In and Adj-RIB-Out

              Adj-RIB-In and Adj-RIB-Out are "views" of routing
              information from the perspective of individual peer

              The Adj-RIB-In contains information advertised to the DUT
              by a specific peer.  The Adj-RIB-Out contains the
              information the DUT will advertise to the peer.

              See RFC 1771[3]. 1771[1].



   Measurement Units: Number of route instances

   See Also: Route, BGP Route, Route Instance, Loc-RIB, FIB

Berkowitz, et al            Expires: August 2002                      9

3.1.2 Loc-RIB

              The Loc-RIB contains the set of best routes selected from
              the various Adj-RIBs, after applying local policies and
              the BGP route selection algorithm.

              The separation implied between the various RIBs is
              logical. It does not necessarily follow that these RIBs
              are distinct and separate entities in any given

              Specifying the RIB is important because the types and
              relative proportions of routes in it can affect the
              convergence efficiency.

              Types of routes can include internal BGP, external
              BGP,interface, static and IGP routes.


   Measurement Units: Number of route instances.

   See Also:  Route, BGP Route, Route Instance, Adj-RIB-in,

3.2 Policy

              Policy is "the ability to define conditions for accepting,
              rejecting, and modifying routes received in

              RFC 1771 [3] [1] further constrains policy to be within the
              hop-by-hop routing paradigm. Policy is implemented using
              filters and associated policy actions.  Many AS's use
              formulate and document their policies using the Routing
              Policy Specification Language (RPSL) [8] to
              document their policies [6] and then
              automatically generate configurations for the BGP
              processes in their routers. routers from the RPSL specifications.

   Measurement Units: Number of policies; length of policies


   See Also: Policy Information Base.

Berkowitz, et al            Expires: August 2002                      10

3.3 Policy Information Base

              A policy information base is the set of incoming and
              outgoing policies.

              All references to the phase of the BGP selection process
              below are made with respect to RFC 1771 [3] [1] definition of
              these phases.

              Incoming policies are applied in Phase 1 of the BGP
              selection process [3] [1] to the Adj-RIB-In routes to set the
              metric for the Phase 2 decision process.  Outgoing
              Policies are applied in Phase 3 of the BGP process to the
              Adj-RIB-Out routes preceding route (prefix and path
              attribute tuple) announcements to a specific peer.

              Policies in the Policy Information Base have matching and
              action conditions.  Common information to match include
              route prefixes, AS paths, communities, etc.  The action on
              match may be to drop the update and not pass it to the
              Loc-RIB, or to modify the update in some way, such as
              changing local preference (on input) or MED (on output),
              adding or deleting communities, prepending the current AS
              in the AS path, etc.

              The amount of policy processing (both in terms of route
              maps and filter/access lists) will impact the convergence
              time and properties of the distributed BGP algorithm. The
              amount of policy processing may vary from a simple policy
              which accepts all routes and sends all routes to complex
              policy with a substantial fraction of the prefixes being
              filtered by filter/access lists.

   Measurement Units: Number and length of policies


   See Also:

Berkowitz, et al            Expires: August 2002                      11

3.4 The Forwarding Information Base (FIB)

              As according to the definition in Appendix B of [4]:
              "The table containing the information necessary to forward
              IP Datagrams is called the Forwarding Information Base.
              At minimum, this contains the interface identifier and
              next hop information for each reachable destination
              network prefix."

              The forwarding information base describes a database
              indexing network prefixes versus router port identifiers.

              The forwarding information base is distinct from the
              "routing table" (the Routing Information Base or RIB),
              which holds all routing information received from routing
              peers. The Forwarding Information Base is generated from
              the RIB.
              The FIB is referred to in [3] as well as [5] but not
              defined in either. For the purposes of this document, the FIB is
              effectively the subset of the RIB used by the forwarding
              plane to make per-packet forwarding decisions.


              Most current implementations have full, non-cached FIBs
              per router interface. All the route computation and
              convergence occurs before a route is entries are downloaded into a

   Measurement Units: units: N.A.


   See Also: Route, RIB

4. Components and characteristics of Routing information

4.1 (Network) Prefix

              A destination address specified in CIDR format. Expressed
              as prefix/length. The definition in [5] is
              "A network prefix is . . . a contiguous set of bits at the
              more significant end of the address that defines a set of
              systems; host numbers select among those systems."
              (This definition is taken directly from section,
              "Classless Inter Domain Routing (CIDR)", in [3].)

              In the CIDR context, the network prefix is expressed as a portion the network
              component of an IP address
              followed by the associated mask such as 10/8. address.

   Measurement Units: N.A.


   See Also

Berkowitz, et al            Expires: August 2002                      12

4.2 Route

   Definition: Network Prefix Length

              The network prefix length is the number of bits used to
              define the network prefix.

              A common alternative to using a bit-wise mask to
              communicate this component is the use of "slash (/)
              notation." Slash notation binds the notion of network
              prefix length (see 4.2) in bits to an IP address. E.g.,
     indicates the network component of this
              IPv4 address is 17 bits wide. Similar notation is used for
              IPv6 network prefixes e.g. :FF02:20::/24

              When referring to groups of addresses, the network prefix
              length is often used as a means of describing groups of
              addresses as an equivalence class.  For example,
              'one hundred /16 addresses' refers to 100 addresses whose
              network prefix length is 16 bits.

   Measurement units: bits


   See Also:  network prefix

4.3 Route

              In general, a 'route' is the n-tuple
              <prefix, nexthop[, other non-routing protocol attributes]>
              A route is not end-to-end, but is defined with respect to
              a specific next hop that will move traffic closer to the
              destination defined by the prefix. In this usage, a route
              is the basic unit of information about a target
              destination distilled from routing protocols.

              This term refers to the concept of a route common to all
              routing protocols. With reference to the definition above,
              typical non-routing-protocol attributes would be
              associated with diffserv or traffic engineering.

   Measurement Units: N.A.

   Issues: None.

   See Also: BGP route


4.4 BGP Route

              A BGP route is an n-tuple
              <prefix, nexthop, aspath ASpath [, other BGP attributes]>.

              BGP Attributes, such as Nexthop is one type of attribute.  Attributes or AS path are defined in
              RFC 1771[3], 1771[1], where they are known as Path Attributes, and
              are the qualifying data that accompanies a prefix the network
              prefixes in a BGP route. route UPDATE message. (An UPDATE message
              may contain multiple prefixes that share a common set of

              From RFC 1771: " For purposes of this protocol a route is
              defined as a unit of information that pairs a destination
              with the attributes of a path to that destination... A
              variable length sequence of path attributes is present in
              every UPDATE.  Each path attribute is a triple
              <attribute type, attribute length, attribute value>
              of variable length."

   Measurement Units: N.A.


   See Also: Route, prefix, Adj-RIB-in.

Berkowitz, et al            Expires: August 2002                      13

4.4 Adj-RIB-in, NLRI.

4.5 BGP Route Instance Attributes and BGP Timers

   The definitions in this section refer to items that are originally
   defined in RFC 1771 [1] and are repeated here for convenience, and to
   allow for some discussion beyond the definitions in RFC 1771.

4.5.1 Network Level Reachability Information (NLRI)

              A single occurrence
              The NRLI consists of a route sent by a one or more network prefixes that
              share all other BGP Peer for a
              particular prefix. When a router has multiple peers from
              which it accepts routes, routes path attributes and are distributed in
              the update portion (as opposed to the same prefix may be
              received from several peers. This is then an example unfeasible routes
              portion) of
              multiple route instances. a BGP UPDATE message.

              Each route instance prefix in the NLRI is associated combined with a specific peer. A
              specific route instance may be rejected by the   (common)
              path attributes in the UPDATE message to form a BGP
              selection algorithm due route.
              The NLRI encapsulates a set of destinations to local policy. which
              packets can be routed (from this point in the network)
              along a common route described by the path attributes.

   Measurement Units: number of route instances N.A.


   See Also:  Route Packing, Network Prefix, BGP Route, NLRI

4.5.2 MinRouteAdvertisementInterval (MRAI)

                (Paraphrased from 1771[1]) The number MRAI timer determines the
              minimum time between advertisements of route instances in routes to a
              particular destination (prefix) from a single BGP device.
              The timer is applied on a pre-prefix basis, although the Adj-RIB-in bases will
              vary based
              timer is set on a per BGP device basis.

              Given that a BGP instance may manage in excess of 100,000
              routes, RFC 1771 allows for a degree of optimization in
              order to limit the function number of timers needed. The NRAI does
              not apply to be performed by a router. An
              inter-provider router, located routes received from BGP speakers in the default free zone
              will likely receive more route same
              AS or to explicit withdrawals.

              RFC 1771 also recommends that random jitter is applied to
              MRAI in an attempt to avoid synchronization effects
              between the BGP instances than in a provider
              edge router, located closer network.

              In this document we define RIB convergence by measuring
              the time an NRAI is advertised to the end-users of DUT to the
              network. time it
              is advertised from the DUT.  Clearly any delay inserted by
              the MRAI will have a significant effect on this

   Measurement Units: seconds.


   See Also:

4.5 Active Route  NLRI, BGP route

4.5.3 MinASOriginationInterval (MAOI)

              Route for which there
              The MAOI specifies the minimum interval between
              advertisements of locally originated routes from this BGP

              Random jitter is a FIB entry corresponding applied to MAOI in an attempt to avoid
              synchronization effects between BGP instances in a
              RIB entry.


   Measurement Units: seconds


   See also: RIB. Also: MRAI, BGP route

4.6 Unique Route Instance

              A unique route instance is a prefix for which there is just one
              route instance single occurrence of a route sent by
              a BGP Peer for a particular prefix. When a router has
              multiple peers from which it accepts routes, routes to the
              same prefix may be received from several peers. This is
              then an example of multiple route instances.

              Each route instance is associated with a specific peer.
              The BGP selection algorithm may reject a specific route
              instance due to local policy.

   Measurement Units: Number of route instances

              The number of route instances in the Adj-RIB-in bases will
              vary based on the function to be performed by a router. An
              inter-provider router, located in the default free zone
              will likely receive more route instances than a provider
              edge router, located closer to the end-users of the

   See Also:

4.7 Active Route

              Route for which there is a FIB entry corresponding to a
              RIB entry.


   Measurement Units: Number of routes.


   See also: RIB.

4.8 Unique Route

              A unique route is a prefix for which there is just one
              route instance across all Adj-Ribs-In.


   Measurement Units: N.A.


   See Also: route, route instance

Berkowitz, et al            Expires: August 2002                      14


4.9 Non-Unique Route

              A Non-unique route is a prefix for which there is at least
              one other route in a set including more than one Adj-RIB-


   Measurement Units: N.A.


   See Also: route, route instance, unique active route.

4.8 Route Packing

4.10BGP UPDATE message

              Number of route prefixes that exist in a single Routing
              An UPDATE Message either as updates (additions or
              modifications) or withdrawals.

              In general, a routing protocol update MAY contain more
              than one prefix.  In BGP, message is an advertisement of a single UPDATE MAY contain NLRI,
              possibly containing multiple prefixes with identical attributes. Protocols
              that do not support such a concept implicitly have a Route
              Packing prefixes, and multiple
              withdrawals of 1. unfeasible routes.  See RFC 1771 ([1]) for


   Measurement Units: N.A.


   See Also: route, route instance,

4.11Characterization of sets of update train.

4.9 Route Mixture

              A characterization messages

   This section contains a sequence of definitions that build up to the routes in
   definition of an Update Train, a collection concept originally introduced by
   Jain and Routhier [11].  This is a formalization of
              routes.  Two profiles are used the sort of test
   stimulus that is expected as input to characterize a
              collection DUT running BGP. This data
   could be a well-characterized, ordered and timed set of routes hand-crafted
   BGP UPDATE packets.  It could just as well be a set of BGP UPDATE
   packets that have been captured from a live router.

   Characterization of route mixture:
              -  The distribution mixtures and Update Trains is an open area
   of prefix lengths in the collection
              - research.  The clustering of prefixes around particular branches
                 the tree which can be used to represent the totality question of possible prefixes.

              Route mixtures are used to simulate interest for this work is
   the normal pattern identification of
              prefix distribution suitable Update Trains, modeled or taken from
   live traces that reflect realistic sequences of UPDATEs and their

4.11.1 Route Packing

              Route packing is seen in real world the number of route
              tables and update trains.  As can be seen prefixes accommodated
              in the analyses a single Routing Protocol UPDATE Message either as
              updates (additions or modifications) or withdrawals.

              In general, a routing protocol update may contain more
              than one prefix.  In BGP, a single UPDATE may contain two
              sets of BGP tables multiple network prefixes: one set of additions
              and traffic (e.g. [12], [13] updates with identical attributes (the NLRI) and [14]), the
              characterizations one
              set of the tables in the routers and the
              network traffic appear to be different which has unfeasible routes to be

Berkowitz, et al            Expires: August 2002                      15
              taken into account in designing test stimuli appropriate
              for particular types of testing. withdrawn.

   Measurement Units:
              Length profile:     Probability distribution function on
                                  possible discrete values of prefix
                                  length (i.e 0-32 for IPv4)
              Clustering profile: Probability distribution function on
                                  'distance' between successive prefixes
                                  which is a function of the prefix
                                  lengths and the separation of the
               Number of the address tree on which
                                  they lie (FFS) prefixes.


   See Also:

4.10 Update Train route, BGP route, route instance, update train, NLRI.

4.11.2 Route Mixture

              A collection of routes such as an NLRI, a set of Routing Protocol UPDATE messages, containing one
              messages or more a RIB.

              A route prefixes, which an external router desires
              to send to the DUT.  When there mixture is more than one prefix in the set, input data for the multiple updates (including withdrawals) may
              be sent benchmark.  The
              particular route mixture used as individual BGP UPDATE packets, or input must be selected to
              suit the question being asked of the benchmark.
              Data containing simple route mixtures, such as one or
              more BGP packets with multiple 100,000 /32
              routes packed (q.v.) into
              them.  If multiple UPDATE packets make up a train, might test the
              time spacing performance limits of the packets needs to be specified.

              The more individual UPDATE packets BGP

              Using live data, or input that are sent, simulates live data, should
              improve understanding of how the more
              TCP and BGP header processing device will operate
              in a live network. The data for this kind of test must be imposed on
              route mixtures that model the
              receiving router patterns of arriving control
              traffic in the live Internet.

              To accomplish that kind of modeling it is necessary to
              identify the DUT. key parameters that characterize a live
              Internet route mixture.  The parameters and how they
              interact is an open research problem.  However, we
              identify the following as affecting the route mixture:
              -  Path length distribution
              -  Attribute distribution
              -  Prefix distribution
              -  Packet packing
              -  Probability density function of inter-arrival times of
                 UPDATES                Each of the items above is more
              complex than a singlenumber.  For example, one could
              consider the distribution of prefixes by AS or
              distribution of prefixes by length.

   Measurement Units: Probability density functions


   See Also:  NLRI, RIB.

4.11.3 Update Train

              An update train may be
              caused by is a variety set of network conditions.  For example,
              an update train could be caused Routing Protocol UPDATE
              messages sent by an influx of UPDATES
              from different peers that have been received and moved a router to
              RIB-out or caused a BGP peer.

              The arrival pattern of UPDATEs can be influenced by many
              things, including TCP parameters, hold-down timers, BGP
              header processing, a peer coming up and advertising its
              routes, or by multiple peers
              sending at the same time.  Network conditions such as a
              local or remote peer flapping a link.
              Other causes are, of course, possible.

              The time intervals between link can also affect the
              arrival pattern.

   Measurement units:
              Probability density function for the inter-arrival times
              of UPDATE packets in a train
              may vary from essentially zero when the packets follow
              each other as closely as possible up train.

              Characterizing the profiles of real world UPDATE trains is
              a matter for future research.  In order to generate
              realistic UPDATE trains as test stimuli a situation where
              the time between packets exceeds MIN_ADVERT_TIME.  Once formal
              mathematical scheme or a proven heuristic is needed to
              drive the packets are so widely spaced in time they can be
              treated separately as single packet update selection of prefixes. Whatever mechanism is
              selected it must generate Update trains since
              the router will be guaranteed that have similar
              characteristics to those measured from live routers.

   See Also:  Route Mixture, MRAI, MAOI

4.11.4 Randomness in Update Trains

   As we have propagated any
              adverts resulting seen from the previous UPDATE when the next
              one arrives (a slightly sweeping statement, but if adverts
              are arriving singly it would be sections, an update train used as a very inefficient router
   test stimulus has a considerable number of parameters that was unable can be
   varied, to keep up!)

   Measurement units:

Berkowitz, et al            Expires: August 2002                      16
              Number of prefixes and UPDATE packets in the train,
              time spacing between packets in train


   See Also:

4.10.1 Random Update Train

              An Update Train which contains:
              - a greater or lesser extent, randomly and independently.

   A random number of prefixes, selected according to Update Train will contain:

   -  A random distribution from the possible updates and
                 withdrawals (depending on what routes are currently
                 installed in the route under test), distributed mixture randomized across
      -  A random number of UPDATE packets as both NLRIs
      - updates and
                 withdraws, in withdrawals
      -  A random order which does not favor any of the
                 well-known algorithms used to manage the route
                 databases in BGP process, and specified for prefixes
      -  A random distribution inter-arrival times of time spacings between
                 deliveries taken from the interval
                 [0, MIN_ADVERT_TIME].

   Discussion: UPDATEs
      and possibly across other variables.

   This is intended to simulate the unpredictable asynchronous nature of
   the network, whereby UPDATE packets may have arbitrary contents and
   be delivered at random times.  A fully random update train can be considered to
              be a worst case in some senses and should stress parts of
              the software:

   It may be desirable to test a router with
              less random cases, such as a not quite random update train
              in which everything is random except important that the UPDATES are
              delivered as closely spaced as possible in time.

              The distribution used data set be randomized sufficiently to control the selection of prefixes
              is a 'route mixture' (q.v.)
              When a router is used in a network of routers all from the
              same vendor, avoid
   favoring one vendor's implementation over another's.
   Specifically, the distribution of prefixes emitted from a
              BGP router may well could be structured in a way which
              particularly suits to
   favor the internal organization of the routes in the databases.  It may be useful to test routers with
              update trains organized in this way as a closer emulation
              of the real world.  One might expect that the input of a particular
   vendor's router would be best suited databases. This is to the ordering
              emitted by a similar router.

   Measurement Units: See Update Train


   See also:

Berkowitz, et al            Expires: August 2002                      17

4.11 Route be avoided.

4.12Route Flap

              RIPE 210 [9] [7] define a route flap as "the announcement and
              withdrawal of prefixes."  For our purposes we define a
              route flap as the rapid withdrawal/announcement or
              announcement/withdrawal of a prefix in the Adj-RIB-in. A
              route flap is not a problem until a route is flapped
              several times in close succession. This causes negative
              repercussions throughout the internet.

              Route flapping can be considered a special and
              pathological case of update trains. A practical
              interpretation of what may be considered excessively rapid
              is the RIPE recommendation of "four flaps in a row". See
              Section 6.1.5 on flap damping for further discussion.

   Measurement units: Flapping events per unit time.

              Specific Flap events can be found in Section 5.1Route
              Change Events.  A bench-marker should use a mixture of
              different route change events in testing.

   See Also: Route change events, flap damping, packet train

5. Route Changes and Convergence

   The following two definitions are central to the benchmarking of
   external routing convergence, and so are singled out for more
   extensive discussion.

5.1 Route Change Events

   A taxonomy characterizing routing information changes seen in
   operational networks is proposed in [6] [4] as well as Labovitz et al[7]. al[5].
   These papers describe BGP protocol-centric events, and event
   sequences in the course of an analysis of network behavior. The
   terminology in the two papers categorizes similar but slightly
   different behaviors with some overlap. We would like to apply these
   taxonomies to categorize the tests under definition where possible,
   because these tests must tie in to phenomena that arise in actual
   networks. We avail ourselves of, or may extend, this terminology as
   necessary for this purpose.

   A route can be changed implicitly by replacing it with another route
   or explicitly by withdrawal followed by the introduction of a new
   route. In either case the change may be an actual change, no change,
   or a duplicate. The notation and definition of individual
   categorizable route change events is adopted from [7] [5] and given

Berkowitz, et al            Expires: August 2002                      18

   a) AADiff: Implicit withdrawal of a route and replacement by a route
      different in some path attribute.
   b) AADup: Implicit withdrawal of a route and replacement by route
      that is identical in all path attributes.
   c) WADiff: Explicit withdrawal of a route and replacement by a
      different route.
   d) WADup: Explicit withdrawal of a route and replacement by a route
      that is identical in all path attributes.

   To apply this taxonomy in the benchmarking context, we need both
   terms to describe the sequence of events from the update train
   perspective, as listed above, as well as and event indications in the time
   domain so as to be able to measure activity from the perspective of
   the DUT. With this in mind, we incorporate and extend the definitions
   of [7] [5] to the following:

   a) Tup (TDx): Route advertised to the DUT by Test Device x
   b) Tdown(TDx): Route being withdrawn by Device x
   c) Tupinit(TDx): The initial announcement of a route to a unique
   d) TWF(TDx): Route fail over after an explicit withdrawal.

   But we need to take this a step further. Each of these events can
   involve a single route, a "short" packet train, or a "full" routing
   table. We further extend the notation to indicate how many routes are
   conveyed by the events above:

   a) Tup(1,TDx) means Device x sends 1 route
   b) Tup(S,TDx) means Device x sends a train, S, of routes
   c) Tup(DFT,TDx) means Device x sends an approximation of a full
      default-free table.

   The basic criterion for selecting a "better" route is the final
   tiebreaker defined in RFC1771, the router ID. As a consequence, this
   memorandum uses the following descriptor events, which are routes
   selected by the BGP selection process rather than simple updates:

   a) Tbest   -- The current best path.
   b) Tbetter -- Advertise a path that is better than Tbest.
   c) Tworse  -- Advertise a path that is worse than Tbest.

5.2 Device Convergence

   Definition: in the Control Plane

              A router routing device is said to have converged onto a route advertised have converged at the point in
              time when the DUT has performed all actions in the control
              plane needed to react to it, given that changes in topology in the route is
              context of the test condition.

              For example, when considering BGP convergence, a change
              that alters the best route instance for a prefix(if multiple choices exist for that prefix), single prefix at
              a router would be deemed to have converged when this route
              is advertised to its downstream peers.

Berkowitz, et al            Expires: August 2002                      19
   Discussion:  Similarly, OSPF
              convergence concludes when SPF calculations have been
              performed and the required link states advertised onwards.

              The convergence process process, in general, can be subdivided
              into three distinct phases:
              -  convergence across the entire Internet,
              -  convergence within an Autonomous System,
              -  convergence with respect to a single router. device.

              Convergence with respect to a single router device can be
              -  convergence with regard to data forwarding process(es)
              -  convergence with regard to the routing process(es), the
                 focus of this document
              - document.

              It is the latter, convergence with regard to data the routing
              process, that we describe in this and the methodology

              Because we are trying to benchmark the routing protocol
              performance which is only a part of the device overall,
              this definition is intended (so far as is possible) to
              exclude any additional time such as is needed to download
              and install the forwarding process(es). information base in the data
              plane.  This definition should be usable for different
              families of protocols.

              It is of key importance to benchmark the performance of
              each phase of convergence separately before proceeding to
              a composite characterization of routing convergence, where
              implementation- specific
              implementation-specific dependencies are allowed to

              The preferred route instance time resolution needed to measure the device
              convergence depends to some extent on the types of the
              interfaces on the router.  For modern routers with gigabit
              or faster interfaces, an individual UPDATE may be
              processed and re-advertised in very much less than a
              millisecond so that time measurements must be unambiguous during
              test setup/definition. made to a
              resolution of hundreds to tens of microseconds or better.

   Measurement Units: N.A.
              Time period.


   See Also:

6. BGP Operation Events

   The BGP speaker process(es) in a router restart device restarts completely, for
   example, because of operator intervention or a power failure, or a
   fails partially because a TCP session has terminated for a particular
   link.  Until recently the BGP process would have to readvertise re-advertise all
   relevant routes on reestablished links potentially triggering updates
   across the network.  Recent work is focused on limiting the volume of
   updates generated by short term outages by providing due to operational events and the amount of processing
   resulting from these events:  This work includes soft refresh[12], a
   graceful restart mechanism [17]. [13] and cooperative route filtering

6.1 Hard reset

              An event which triggers a complete reinitialization re-initialization of
              the routing tables on one or more BGP sessions, resulting
              in exchange of a large number of UPDATEs full routing table on one or more links
              to the router.


   Measurement Units: N/A


   See Also:

Berkowitz, et al            Expires: August 2002                      20

6.2 Soft reset

              An event which results in a complete or partial restart of
              the BGP session(s) on a router, BGP device, but which avoids the
              exchange of a large number of UPDATEs full table by maintaining state across the


   Measurement Units: N/A


   See Also:

7. Factors that impact the performance of the convergence process

   While this is not a complete list, all of the items discussed below
   have a significant affect on BGP convergence.  Not all of them can be
   addressed in the baseline measurements described in this document.

7.1 General factors affecting BGP device convergence

   These factors are conditions of testing external to the router Device
   Under Test (DUT).

7.1.1 Number of peers

   As the number of peers increases, the BGP route selection algorithm
   is increasingly exercised. In addition, the phasing and frequency of
   updates from the various peers will have an increasingly marked
   effect on the convergence process on a router as the number of peers
   grows. Increasing the number of peers also increases the processing
   workload for TCP and BGP keepalives.

7.1.2 Number of routes per peer

   The number of routes per BGP peer is an obvious stressor to the
   convergence process. The number, and relative proportion, of multiple
   route instances and distinct routes being added or withdrawn by each
   peer will affect the convergence process, as will the mix of
   overlapping route instances, and IGP routes.

7.1.3 Policy processing/reconfiguration

   The number of routes and attributes being filtered, and set, as a
   fraction of the target route table size is another parameter that
   will affect BGP convergence.

   Extreme examples are
   -  Minimal Policy: receive all, send all,
   -  Extensive policy: up to 100% of the total routes have applicable

Berkowitz, et al            Expires: August 2002                      21

7.1.4 Interactions with other protocols.

   There are interactions in the form of precedence, synchronization,
   duplication and the addition of timers, and route selection criteria.
   Ultimately, understanding BGP4 convergence must include understanding
   of the interactions with both the IGPs and the protocols associated
   with the physical media, such as Ethernet, SONET, DWDM.

7.1.5 Flap Damping

   A router can use flap damping to  respond to route flapping.   Use of
   flap damping is not mandatory, so the decision to enable the feature,
   and to change parameters associated with it, can be considered a
   matter of routing policy.

   The timers are defined by RFC 2439 [4] [2] and discussed in RIPE-210 [9]. RIPE-229 [7].
   If this feature is in effect, it requires that the router device keep
   additional state to carry out the damping, which can have a direct
   impact on the control plane due to increased processing.  In
   addition, flap damping may delay the arrival of real changes in a
   route, and affect convergence times

7.1.6 Churn

   In theory, a BGP router device could receive a set of updates that
   completely defined the Internet, and could remain in a steady state,
   only sending appropriate KeepAlives. keepalives.  In practice, the Internet will
   always be changing.

   Churn refers to control plane processor activity caused by
   announcements received and sent by the router.  It does not include
   keepalives and TCP processing.

   Churn is caused by both normal and pathological events.  For example,
   if an interface of the local router goes down and the associated
   prefix is withdrawn, that withdrawal is a normal activity, although
   it contributes to churn.  If the local router device receives a withdrawal
   of a route it already advertises, or an announcement of a route it
   did not previously know, and readvertises re-advertises this information, again
   these are normal constituents of churn. Routine updates can range
   from single announcement or withdrawals, to announcements of an
   entire default-free table.  The latter is completely reasonable as an
   initialization condition.

   Flapping routes are a pathological contributor to churn, as is MED
   oscillation [11]. [16].  The goal of flap damping is to reduce the
   contribution of flapping to churn.

   The effect of churn on overall convergence depends on the processing
   power available to the control plane, and whether the same
   processor(s) are used for forwarding and for control.

7.2 Implementation-specific and other factors affecting BGP convergence

Berkowitz, et al            Expires: August 2002                      22

   These factors are conditions of testing internal to the router Device Under
   Test (DUT), although they may affect its interactions with test

7.2.1 Forwarded  traffic

   The presence of actual traffic in the router device may stress the control
   path in some fashion if both the offered load due to data and the
   control traffic (FIB updates and downloads as a consequence of
   flaps)are excessive. The addition of data traffic presents a more
   accurate reflection of realistic operating scenarios than if only
   control traffic is present.

7.2.2 Timers

   Settings of delay and hold-down timers at the link level as well as
   for BGP4, can introduce or ameliorate delays.  As part of a test
   report, all relevant timers should be reported if they use non-
   default value.  Also, any variation in standard behavior, such as
   overriding TCP slow start, should be documented.

7.2.3 TCP parameters underlying BGP transport

   Since all BGP traffic and interactions occur over TCP, all relevant
   parameters characterizing the TCP sessions should be provided: eg Max
   Slow start, max window size, Maximum maximum segment size, or timers.

7.2.4 Authentication

   Authentication in BGP is currently done using the TCP MD5 Signature
   Option [10]. [8].  The processing of the MD5 hash, particularly in routers devices
   with a large number of BGP peers and a large amount of update traffic
   can have an impact on the control plane of the router. device.

8. Security Considerations

   The document explicitly considers authentication as a performance-
   affecting feature, but does not consider the overall security of the
   routing system.

Berkowitz, et al            Expires: August 2002                      23

9. References


     [1]            Bradner, S., "The Internet Standards Process --
                    Revision 3", BCP 9, RFC 2026, October 1996

     [2]            Bradner, S., "Key words for use in RFCs to
                    Indicate Requirement Levels", BCP 14, RFC 2119,
                    March 1997

     [3]            Rekhter, Y. and Li, T., "A Border Gateway
                    Protocol 4 (BGP-4)", RFC 1771, . March 1995.


     [2]            Villamizar, C., Chandra, R. and Govindan, R.,
                    "BGP Route Flap Damping", RFC 2439,
                    November 1998."


     [3]            Baker, F.,"Requirements for IP Version 4
                    Routers", RFC 1812. June 1995.


     [4]            Ahuja, A., Jahanian, F., Bose, A. and Labovitz,
                    "An Experimental Study of Delayed Internet
                    Routing Convergence", RIPE 37 - Routing WG.


     [5]            Labovitz, C., Malan, G.R. and Jahanian, F.,
                    "Origins of Internet Routing Instability,"
                    Infocom 99,

     [8] 99.

     [6]            Alaettinoglu, C., Villamizar, C., Gerich, E.,
                    Kessens, D.,  Meyer, D., Bates, T., Karrenberg,
                    D. and Terpstra, M., "Routing Policy
                    Specification Language (RPSL)", RFC 2622, June


     [7]            Barber, T., Doran, S., Karrenberg, D., Panigl,
                    C., Schmitz, J., "RIPE Routing-WG Recommendation
                    for coordinated route-flap damping parameters",
                    RIPE 210,

     [10] 210.

     [8]            Heffernan, A., "Protection of BGP Sessions via
                    the TCP MD5 Signature Option", RFC 2385, August

     [9]            Juniper Networks,"Junos(tm) Internet Software
                    Configuration Guide Routing and Routing
                    Protocols, Release 4.2"
                    September 2000 (and other releases).

     [10]           Rosen, E. and Rekhter, Y., "BGP/MPLS VPNs", RFC
                    2547, March 1999.

     [11]           McPhersonm,           Jain, R. and Routhier, S.A., "Packet trains --
                    measurement and a new model for computer network
                    traffic," IEEE Journal on Selected Areas in
                    Communication, 4(6)September 1986.


     [12]           Chen, E., "Route Refresh for BGP-4", RFC 2918,
                    September 2000.

     [13]           Ramachandra, S., Rekhter, Y., Fernando, R.,
                    Scudder, J.G. and Chen, E.,
                    "Graceful Restart Mechanism for BGP",
                    draft-ietf-idr-restart-02.txt, January 2002,
                    work in progress.

     [14]           Chen, E. and Rekhter, Y, "Cooperative Route
                    Filtering Capability for BGP-4",
                    draft-ietf-idr-route-filter-05.txt, January 2002,
                    work in progress.

     [15]           T. Anderson et al. "Requirements for Separation
                    of IP Control and Forwarding",
                    February 2002, work in progress.

     [16]           McPherson, Gill, Walton, Retana,  "BGP Persistent
                    Route Oscillation Condition", <draft-
                    ietf-idr-route-oscillation-00.txt>, Work
                    February 2002, work In

     [12] progress.

     [17]           Bates, T., "The CIDR Report",
                    Internet statistics relevant to inter-domain
                    routing updated daily

Berkowitz, et al            Expires: August 2002                      24
     [13] daily.

     [18]           Smith, P. (designer), APNIC Routing Table
                    Statistics derived from a daily analysis of a
                    core router in Japan

     [14] Japan.

     [19]           Huston, G., Telstra BGP table statistics,
                    Statistics derived daily from the BGP tables of
                    Telstra and other AS's routers

     [15]           Juniper Networks,"Junos(tm) routers.

   For Internet Software
                    Configuration Guide Routing and Routing
                    Protocols, Release 4.2"
                    September 2000 (and other releases).

     [16]           Rosen, E. and Rekhter, Y., "BGP/MPLS VPNs", RFC
                    2547, March 1999

     [17]           Ramachandra, Draft consistency purposes only

     [20]           Bradner, S., Rekhter, Y., Fernando, R.,
                    Scudder, J.G. and Chen, E.,
                    "Graceful Restart Mechanism for BGP",
                    draft-ietf-idr-restart-02.txt, January 2002, work
                    in progress. "The Internet Standards Process --
                    Revision 3", BCP 9, RFC 2026, October 1996

   Thanks to Francis Ovenden and Elwyn Davies for review and Abha Ahuja for
   encouragement. Much appreciation to Jeff Haas, Matt Richardson, and
   Shane Wright at Nexthop for comments and input. Debby Stopp and Nick
   Ambrose contributed the concept of route packing.

11.Author's Addresses

   Howard Berkowitz
   Gett Communications
   5012 S. 25th St
   Arlington VA 22206
   Phone: +1 703 998-5819
   Fax:   +1 703 998-5058

   Elwyn Davies
   Nortel Networks
   London Road
   Harlow, Essex CM17 9NA
   Phone: +44-1279-405498

   Susan Hares
   Nexthop Technologies
   517 W. William

Berkowitz, et al            Expires: August 2002                      25
   Ann Arbor, Mi 48103

   Padma Krishnaswamy

   Marianne Lepp
   Juniper Networks
   51 Sawyer Road
   Waltham, MA 02453
   Phone: 617 645 9019

   Alvaro Retana
   Cisco Systems, Inc.
   7025 Kit Creek Rd.
   Research Triangle Park, NC 27709

Full Copyright Statement

   Copyright (C) The Internet Society (2002). All Rights Reserved.

   This document and translations of it may be copied and furnished to
   others, and derivative works that comment on or otherwise explain it
   or assist in its implmentation implementation may be prepared, copied, published
   and distributed, in whole or in part, without restriction of any
   kind, provided that the above copyright notice and this paragraph are
   included on all such copies and derivative works. However, this
   document itself may not be modified in any way, such as by removing
   the copyright notice or references to the Internet Society or other
   Internet organizations, except as needed for the purpose of
   developing Internet standards in which case the procedures for
   copyrights defined in the Internet Standards process must be
   followed, or as required to translate it into languages other than

   The limited permissions granted above are perpetual and will not be
   revoked by the Internet Society or its successors or assigns.

   This document and the information contained herein is provided on an

Berkowitz, et al            Expires: August 2002                      26