S/MIME Working Group					Weston Nicolls
INTERNET DRAFT							Ernst & Young LLP						Telenisus Corporation
Expires in six months 						December 1999 					July 2000

		  Implementing Company Classification Policy
			with the S/MIME Security Label
		      <draft-ietf-smime-seclabel-00.txt>
		      <draft-ietf-smime-seclabel-01.txt>

Status of this memo

This document is an Internet-Draft and is in full conformance with all
provisions of Section 10 of [RFC2026].

This document is an Internet-Draft. 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
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.

<<Comments are contained in angle brackets like this.>>

1.   Introduction

This document discusses how company security policy for data classification can
be mapped to the S/MIME classification label.  Actual policies from 3 companies
are used to provide worked examples.

Security labels are an optional security service for S/MIME. A security label is
a set of security information regarding the sensitivity of the content that is
protected by S/MIME encapsulation. A security label can be a bound attribute of included in the original message content,
signed attributes of any SignedData object.  A security label attribute may be
included in either the encrypted body, inner signature, outer signature, or both.  The syntax
and processing rules for security labels are described in [ESS].

The key words 'MUST', 'MUST NOT', 'REQUIRED', 'SHALL', 'SHALL NOT','SHOULD',
'SHOULD NOT', 'RECOMMENDED',  'MAY', and 'OPTIONAL' in this document are to be
interpreted as described in [MUSTSHOULD].

This draft is being discussed on the 'ietf-smime' mailing list.  To join the
list, send a message to <ietf-smime-request@imc.org> with the single word
'subscribe' in the body of the message.  Also, there is a Web site for the
mailing list at <http://www.imc.org/ietf-smime>.

1.1 Information Classification Policies

Information is an asset, but not all information has the same value for a
business. Not all information needs to be protected as strongly as other
information.

Research and development plans, marketing strategies and manufacturing quality
specifications developed and used by a company provide competitive advantage.
This type of information needs stronger protective measures than other
information, which if disclosed or modified, would cause little or no damage to
the company.

Other types of information such as internal organization charts, employee lists
and policies may need little and no protective measures based on value the
organization places on it.

A corporate information classification policy defines how its information assets
are to be protected.  It provides guidance to employees on how to classify
information assets.  It defines how to label and protect an asset based on its
classification and state (e.g. facsimile, electronic transfer, storage,
shipping, etc.).

1.2 Access Control and Security Labels

"Access control" is a means of enforcing authorizations. There are a variety of
access control methods that are based on different types of policies and rely on
different security mechanisms.

-  Rule based access control is based on policies that can be algorithmically
expressed.

-  Identity based access control is based on a policy which applies explicitly
to an individual person or host entity, or to a defined group of such entities.
Once identity has been authenticated, if the identity is verified to be on the
access list, then access is granted.

-  Rank base access control is based on a policy of hierarchical positions in an
organization.  It is based on who you are in the company structure. A rank-based
policy would define what information that the position of Partner or Senior
Consultant could access.

-  Role based access control is based on a policy of hierarchical roles in an organization.
It may or may not be hierarchical.  It is based on who you are in the company.
The role-based policy would define what information that the role of Executive, Vice President Database
Administrator, Network Administrator, Mailroom Clerk or Staff Purchaser could access.

Rule, rank and role-based access control methods can rely on a security label as
the security mechanism to convey the sensitivity or classification of the
information.  When verifying an S/MIME encapsulated message, the sensitivity
information in the messages message's security label can be compared with the recipient's
authorizations to determine if the recipient is allowed to access the protected
content.

An S/MIME security label may be included as an authenticated attribute in the
inner (or only) signature or the outer signature.  The inner signature would be
used for access control decisions related to the plaintext original content,
while he the outer signature would be used for access control decisions related to
the encrypted message.

1.3 User Authorizations

Users need to be granted authorizations to access information that has been
classified by an authority.  The sending and receiving agent need to be able to
securely determine the users user's authorizations for access control processing.

[X.509] and the Internet profile for X.509 certificates [CERTCRL] do not define
the means to represent and convey authorizations in a certificate.

[X.501] defines how to represent authorization in the form of a clearance
attribute.  The clearance attribute identifies the security policy in force to
which a list of possible classifications and security categories relates.

[X.501] also notes two means for binding the clearance to a named entity: an
Attribute Certificate and a Certificate extension field (e.g., within the
subjectDirectoryAttribute extension).

[AC509] defines a profile of X.509 Attribute Certificate (AC) suitable for use
with authorization information within Internet Protocols. One of the defined
attributes is Clearance, which carries clearance (security labeling) information
about the AC owner.  The syntax for Clearance is imported from [X.501].

2. Developed Examples

2.1 Classification Policies

The following describes the information classification policies in effect at 3
companies.

2.1.1 Amoco Corporation

The description for the Amoco information classification policy was taken from
the Amoco Computer Security Guidelines.  Amoco classifies its information assets
based on confidentiality and integrity and defines 3 hierarchical
classifications for each.

Highly Confidential  The confidentiality and integrity polices are
independent, so either or both may be applied to the information.  Amoco also
defines an availability classification for time critical information.

HIGHLY CONFIDENTIAL - Information whose unauthorized disclosure will cause the
company severe financial, legal or reputation damage. Examples:  Certain
acquisitions, bid economics, negotiation strategies.

Confidential

CONFIDENTIAL - Information whose unauthorized disclosure may cause the company
financial, legal, or reputation damage. Examples:  Employee Personnel & Payroll
Files, some interpreted Exploration Data.

General

GENERAL - Information that, because of its personal, technical, or business
sensitivity is restricted for use within the company. Unless otherwise
classified, all information within Amoco is in this category.

Maximum

MAXIMUM - Information whose unauthorized modification and destruction will cause
the company severe financial, legal, or reputation damage.

Medium

MEDIUM - Information whose unauthorized modification and destruction may cause
the company financial, legal, or reputation damage. Examples: Electronic Funds,
Transfer, Payroll, and Commercial Checks

Minimum

MINIMUM - Although an error in this data would be of minimal consequence, this
is still important company information and therefore will require some minimal
controls to ensure a minimal level of assurance that the integrity of the data
is maintained. This applies to all data that is not placed in one of the above
classifications. Examples: Lease Production Data, Expense Data, Financial Data,
and Exploration Data.

CRITICAL - It is important to assess the availability requirements of data,
applications and systems. A business decision will be required to determine the
length of unavailability that can be tolerated prior to expending additional
resources to ensure the information availability that is required. Information
should be labeled "CRITICAL" if it is determined that special procedures should
be used to ensure its' availability.

2.1.2 Caterpillar, Inc.

The description for the Caterpillar information classification policy is taken
from the Caterpillar Information Protection Guidelines.  Caterpillar classifies
its information assets based on confidentiality and defines 4 hierarchical
classifications.

Caterpillar Confidential Red - Provides a significant competitive advantage.
Disclosure would cause severe damage to operations. Relates to or describes a
long-term strategy or critical business plans. Disclosure would cause regulatory
or contractual liability. Disclosure would cause severe damage to our reputation
or the public image. Disclosure would cause a severe loss of market share or the
ability to be first to market. Disclosure would cause a loss of an important
customer, shareholder, or business partner. Disclosure would cause a long-term
or severe drop in stock value. Strong likelihood somebody is seeking to acquire
this information.

Caterpillar Confidential Yellow - Provides a competitive advantage. Disclosure
could cause moderate damage to the company or an individual. Relates to or
describes an important part of the operational direction of the company over
time. Important technical or financial aspects of a product line or a business
unit. Disclosure could cause a loss of Customer or Shareholder confidence.
Disclosure could cause a temporary drop in stock value. A likelihood that
somebody could seek to acquire this information.

Caterpillar Confidential Green - Might provide a business advantage over those
who do not have access to the same information. Might be useful to a competitor.
Not easily identifiable by inspection of a product. Not generally known outside
the company or available from public sources. Generally available internally.
Little competitive interest.

Caterpillar Public - Would not provide a business or competitive advantage.
Routinely made available to interested members of the General Public.  Little or
no competitive interest.

2.1.3 Whirlpool Corporation

The description for the Whirlpool information classification policy is taken
from the Whirlpool Information Protection Policy.  Whirlpool classifies its
information assets based on confidentiality and defines 2 hierarchical
classifications.  The policy states that:

"All information generated by or for Whirlpool, in whatever form, written,
verbal, or electronic, is to be treated as WHIRLPOOL INTERNAL or WHIRLPOOL
CONFIDENTIAL.  Classification of information in either category depends on its
value, the impact of unauthorized disclosure, legal requirements, and the manner
in which it needs to be used by the company.  Some WHIRLPOOL INTERNAL
information may be authorized for public release."

WHIRLPOOL CONFIDENTIAL - A subset of Whirlpool Internal information, the
unauthorized disclosure or compromise of which would likely have an adverse
impact on the companys company's competitive position, tarnish its reputation, or
embarrass an individual.  Examples: Customer, financial, pricing, or personnel
data; merger/acquisition, product, or marketing plans; new product designs,
proprietary processes and systems.

WHIRLPOOL INTERNAL - All forms of proprietary information originated or owned by
Whirlpool, or entrusted to it by others.  Examples: Organization charts,
policies, procedures, phone directories, some types of training materials.

WHIRLPOOL PUBLIC - Information officially released by Whirlpool for widespread
public disclosure. Example: Press releases, public marketing materials,
employment advertising, annual reports, product brochures, the public web site,
etc

The policy also states that privacy markings are allowable. Specifically:

For WHIRLPOOL INTERNAL, additional markings or caveats are option at the
discretion of the information owner.

For WHIRLPOOL CONFIDENTIAL, add additional marking or caveats as necessary to
comply with regulatory or heightened security requirements.  Examples:  MAKE NO
COPIES, THIRD PARTY CONFIDENTIAL, ATTORNEY-CLIENT PRIVILEGED DOCUMENT,
DISTRIBUTION LIMITED TO ____, COVERED BY A NON-ANALYSIS AGREEMENT.

2.2 S/MIME Classification Label Developed Examples

[ESS] defines the ESSSecurityLabel syntax and processing rules.  This section
builds upon those definitions to define detailed example policies.

2.2.1 Security Label Components

The examples are detailed using the various components of the eSSSecurity Label
syntax.

2.2.1.1 Security Policy Identifier

A security policy is a set of criteria for the provision of security services.
The eSSSecurityLabel security-policy-identifier is used to identify the security
policy in force to which the security label relates. It indicates the semantics
of the other security label components.

For the example policies, the following security policy object identifiers are
defined:

<<Need to obtain test

-- S/MIME OIDs.>>

Amoco security-policy-identifier Working Group Object Identifier Registry
id-smime OBJECT IDENTIFIER ::= { iso(1) member-body(2) us(840) rsadsi(113549)
					pkcs(1) pkcs-9(9) 16 }

Caterpillar security-policy-identifier

-- S/MIME Test Securty Policy Arc
id-tsp  OBJECT IDENTIFIER ::= { id-smime 7 }

Whirlpool security-policy-identifier

-- Test Security Policies
id-tsp-TEST-Amoco          OBJECT IDENTIFIER ::= { id-tsp 1 }
id-tsp-TEST-Caterpillar    OBJECT IDENTIFIER ::= { id-tsp 2 }
id-tsp-test-Whirlpool      OBJECT IDENTIFIER ::= { id-tsp 3 }

2.2.1.2 Security Classification

The security classification values and meanings are defined by the governing
company policies. The security-classification values defined are hierarchical
and do not use integers 0 through 5.

Amoco-SecurityClassification ::=  {
  amoco general (6),
  amoco confidential (7),
  amoco highly confidential (8),
  amoco minimum (9),
  amoco medium (10),
  amoco maximum (11) (8) }  (0..ub-integer-options)

Caterpillar-SecurityClassification values ::=  {
  caterpillar public (6),
  caterpillar green (7),
  caterpillar yellow (8),
  caterpillar red (9) }  (0..ub-integer-options)

Whirlpool-SecurityClassification values ::=  {
  whirlpool public (6),
  whirlpool internal (7),
  whirlpool confidential (8) }  (0..ub-integer-options)

2.2.1.3 Privacy Mark

Privacy marks are specified the Whirlpool policy. The policy provides examples
of possible marking markings but other can be defined by users as necessary.  User
specified necessary (though no
guidance is given).  The Whirlpool policy provides the following examples:
MAKE NO COPIES, THIRD PARTY CONFIDENTIAL, ATTORNEY-CLIENT PRIVILEGED DOCUMENT,
DISTRIBUTION LIMITED TO ____, and COVERED BY A NON-ANALYSIS AGREEMENT.

The Amoco policy does not identify any privacy marks are but the classification
labels defined using for availability and integrity would be most appropriately
displayed here.  The CRITICAL, MAXIMUM, MEDIUM, MINIMUM labels are examples of
information classification that are not used for access control.

In general, the following syntax.

<<Need privacy marks should provide brief but clear direction to develop the syntax.  Suggestions?>>
user on how to handle the information.

2.2.1.4 Security Categories

Security categories or caveats are not specified to any of the sample policies.
However, they are used in at least 2 of the companies informally. companies.  Though formal the security
categories are not defined, some proprietary information does need more
granular access control.  A category can be based organizationally or by project
(i.e., Legal defined formally in the security policies, once locally
defined they are formal and are to be enforced. The security categories are
defined when necessary to provide identifiable proprietary information more
granular access control.  A category can be based organizationally or by project
(i.e., Legal Only or Project Vallor).

2.2.1.4.1 Syntax

User specified security categories categories are defined using the following syntax
(taken and expanded from [ESS]).

SecurityCategories ::= SET SIZE (1..ub-security-categories) OF SecurityCategory

ub-security-categories INTEGER ::= 64

SecurityCategory ::= SEQUENCE {
  type	[0] OBJECT IDENTIFIER		     <<need to get a generic SMIME OID>>
  value	[1] ANY DEFINED BY type -- defined by type

When id-securityCategoryValues is present in the SecurityCategory type field, then
the SecurityCategory value field will take the form of SecurityCategoryValues as
follows:

SecurityCategoryValues ::= SEQUENCE OF SecurityCategoryTagGroup

SecurityCategoryTagGroup ::=  SEQUENCE {
  securityCategoryTagGroupName	OBJECT IDENTIFIER,
  securityCategoryTagGroup	SEQUENCE OF SecurityCategoryTag}

SecurityCategoryTag  ::= CHOICE {
  restrictiveBitMap		[0] BIT STRING,
  permissiveBitMap		[1] BIT STRING,
  noAccessControlBitMap	  	[2] BIT STRING,
  restrictiveEnumerated	  	[3] SEQUENCE OF INTEGER,
  permissiveEnumerated 	  	[4] SEQUENCE OF INTEGER,
  noAccessControlEnumerated 	[5] SEQUENCE OF INTEGER}

2.2.1.4.2 Use

An organization will define a securityCategoryTagGroupName OID representing each
unique group of defined security category values (i.e. SecurityCategoryTagGroup)
within their security policy.  An organization can define multiple
SecurityCategoryTagGroups within their security policy. A
SecurityCategoryTagGroup MUST not include multiple definitions for a
SecurityCategoryTag CHOICE.  For example, there MUST not be multiple
permissiveBitMaps defined within a single SecurityCategoryTagGroup.

The Restrictive Bit Map Tag bit string is used to convey a set of non-hierarchical
attributes that apply to the labeled message. A bit is assigned to every security
policy-defined restrictive attribute. Bits corresponding to restrictive attributes
that apply will be set to 1, otherwise bits are set to 0. Access MUST be restricted
to only those entities whose set of attributes is a subset of the attributes for
the receiving end. Security compartments and caveats are examples of restrictive
security attributes.

The Permissive Bit Map Tag bit string is used to convey a set of non-hierarchical
attributes that apply to the labeled message. A bit is assigned to every security
policy-defined permissive attribute. Release markings are examples of permissive
security attributes. Bits corresponding to types or groups of entities that are
granted access to the message are set to 0, all other bits are set to 1. For
example, the label on entities to be available only to members of an
organization's Personnel Office will have the bit assigned to the Personnel
Office set to 0 and all others set to 1. A message can be accepted if the
receiving end belongs to any of the release groups in the release permission list
on the message label.

The No Access Control Bit Map Tag bit string is used to to represent security
category values for which no access control check is required (e.g. originator
controlled data).

The Restrictive Enumerated Tag non-negative integers convey a security level, a
hierarchical security attribute. The higher the value of this attribute the
higher the security level of the labeled message. Each of the integers that
follow represent a non-hierarchical attribute that applies to the labeled
message. This is an alternative to the bit-representation in the Restrictive Bit
Map Tag. It is intended to minimize label length in cases where only a few
attributes out of a large set apply to the message. An example of attributes
enumerated by this tag are compartments. Access MUST be restricted to
only those messages whose set of attributes is a subset of the attributes for the
receiving end.

The Permissive Enumerated Tag non-negative integers convey a security level, a
hierarchical security attribute. The higher the value of this attribute the
higher the security level of the labeled message. Each of the integers that
follow represent a non-hierarchical attribute that applies to the labeled
message. This is an alternative to the bit-representation in the Permissive
Enumerated Tag. It is intended to minimize label length in cases where only a few
attributes out of a large set apply to the message. An example of attributes
enumerated by this tag are defined
using release permissions. A message can be accepted if the following syntax.

<< Need
receiving end belongs to develop any of the syntax.  Suggestions?>> release groups in the release permission list
on the message label.

The No Accesss Control Enumerated Tage non-negative integers convey security category
values for which no access control check is required (e.g. list of country codes).

2.2.1.4.3 Security Category Example

<<need to an at least example for each type with a test securityCategoryTagGroupNam
OID and sample values>>

2.2.2 Attribute Owner Clearance

The security clearance and category authorizations for the user are defined in
the clearance attribute.

2.2.2.1 Amoco User

Clearance  ::=  SEQUENCE {
           policyId  OBJECT IDENTIFIER,  1 2 840 113549 1 9 16 7 1,
           classList ClassList DEFAULT {general},
           securityCategories
                      SET OF SecurityCategory  OPTIONAL
           }

           ClassList  ::=  BIT STRING {
               amoco general              (6),
               amoco confidential         (7),
               amoco highly confidential  (8),
           }

           SecurityCategory ::= SEQUENCE {
               type      [0]  IMPLICIT OBJECT IDENTIFIER,
               value     [1]  ANY DEFINED BY type
           }

2.2.2.1

2.2.2.2 Caterpillar User

Clearance  ::=  SEQUENCE {
           policyId  OBJECT IDENTIFIER,  1 2 840 113549 1 9 16 7 2,
           classList ClassList DEFAULT {general},
           securityCategories
                      SET OF SecurityCategory  OPTIONAL
           }

           ClassList  ::=  BIT STRING {
               caterpillar public              (6),
               caterpillar confidential greeen (7),
               caterpillar confidential yellow (8),
               caterpillar confidential red    (9)
           }

           SecurityCategory ::= SEQUENCE {
               type      [0]  IMPLICIT OBJECT IDENTIFIER,
               value     [1]  ANY DEFINED BY type
           }

2.2.2.1

2.2.2.3 Whirlpool User

Clearance  ::=  SEQUENCE {
           policyId  OBJECT IDENTIFIER,  1 2 840 113549 1 9 16 7 3,
           classList ClassList DEFAULT {general},
           securityCategories
                      SET OF SecurityCategory  OPTIONAL
           }

           ClassList  ::=  BIT STRING {
               whirlpool public        (6),
               whirlpool internal      (7),
               whirlpool confidential  (8),
           }

           SecurityCategory ::= SEQUENCE {
               type      [0]  IMPLICIT OBJECT IDENTIFIER,
               value     [1]  ANY DEFINED BY type
           }

<<need to complete security category with example OIDs and values>>

2.2.3 Additional ESSSecurityLabel Processing Guidance

<<What needs

An implementation issue with the security label is the mapping of the bit string
to be addressed here?>>

<<How does displayable characters.  This is an issue for users who want to develop and
retire their own classifications and categories a regular basis.  Applications
should provide a means for the app know how enterprise to interpret manage these changes.  The practice
of hard coding the values mapping into meaningful text?>>

When originating enveloped data, the agents MUST allow applications is discouraged.

This issue is viewed as local issue for the security label of application vendor, as the
data solution
does not need to be specified.

Upon successful access control processing, interoperable between vendors.

An approach is the agents SHOULD display use of a Security Policy Information File (SPIF)[X.sio].  A
SPIF is a construct that conveys domain-specific security policy information.
It is also a sign object to protect it from unauthorized changes and to
authenticate the
recipient source of the security label policy information.  It contains critical display
information such as the text string for security classifications and security
categories to be displayed to the encrypted data.

<<What is missing?>> user, as well as additional security policy
information.

3. Security Considerations

All security considerations from [CMS] and [ESS] apply to applications that use
procedures described in this document.

A. References

[AC509] Farrell, S., Housley, R., "An Internet AttributeCertificate Attribute Certificate Profile for
Authorization", draft-ietf-pkix-ac509prof-01.txt. draft-ietf-pkix-ac509prof-03.txt.

[CMS] Housley, R., "Cryptographic Message Syntax", RFC 2630.

[ESS] Hoffman, P., Editor, "Enhanced Security Services for S/MIME", RFC 2634.

[MUSTSHOULD] Bradner, S., "Key Words for Use in RFCs to Indicate Requirement
Levels", RFC 2119.

[X.501] "ITU-T Recommendation X.501 : X.501: Information Technology - Open Systems
Interconnection - The Directory: Models", 1993.

[X.509] "ITU-T Recommendation X.509 (1997 E): Information
Technology - Open Systems Interconnection - The Directory: Authentication
Framework", June 1997.

[X.sio] "Information Technology - Security Techniques - Security Information
Objects", IOS/IEC FCD 15816, 1999-11-12.

B. Acknowledgements

I would like to thanks thank Russ Housley for helping me through the process of
developing this document.  I would like to thank John Pawling for his technical
assistance and guidance.  I would also like to thank the good people at (BP)
Amoco, Caterpillar and Whirlpool who allowed me to use their policies as the
real examples that make this document possible.

C. Authors Author's Address

Weston Nicolls
Telenisus Corporation (formerly with Ernst & Young LLP
111 N. Canal St
Chicago, LLP)
1701 Golf Rd
Tower 3, Suite 600
Rolling Meadows, IL 60606
(312) 879-2075
weston.nicolls@ey.com 60008
(847) 871-5086
wnicolls@telenisus.com

D. Open issues: