Network Working Group                                            S. Kent
Request for Comments: 2401 BBN Corp
Obsoletes: 1825 R. Atkinson
Category: Standards Track @Home Network
November 1998

Security Architecture for the Internet Protocol

Status of this Memo

This document specifies an Internet standards track protocol for the
Internet community, and requests discussion and suggestions for
improvements. Please refer to the current edition of the "Internet
Official Protocol Standards" (STD 1) for the standardization state
and status of this protocol. Distribution of this memo is unlimited.

Copyright Notice

Copyright (C) The Internet Society (1998). All Rights Reserved.

Table of Contents

1. Introduction........................................................3
1.1 Summary of Contents of Document..................................3
1.2 Audience.........................................................3
1.3 Related Documents................................................4
2. Design Objectives...................................................4
2.1 Goals/Objectives/Requirements/Problem Description................4
2.2 Caveats and Assumptions..........................................5
3. System Overview.....................................................5
3.1 What IPsec Does..................................................6
3.2 How IPsec Works..................................................6
3.3 Where IPsec May Be Implemented...................................7
4. Security Associations...............................................8
4.1 Definition and Scope.............................................8
4.2 Security Association Functionality..............................10
4.3 Combining Security Associations.................................11
4.4 Security Association Databases..................................13
4.4.1 The Security Policy Database (SPD).........................14
4.4.2 Selectors..................................................17
4.4.3 Security Association Database (SAD)........................21
4.5 Basic Combinations of Security Associations.....................24
4.6 SA and Key Management...........................................26
4.6.1 Manual Techniques..........................................27
4.6.2 Automated SA and Key Management............................27
4.6.3 Locating a Security Gateway................................28
4.7 Security Associations and Multicast.............................29

5. IP Traffic Processing..............................................30
5.1 Outbound IP Traffic Processing..................................30
5.1.1 Selecting and Using an SA or SA Bundle.....................30
5.1.2 Header Construction for Tunnel Mode........................31
5.1.2.1 IPv4 -- Header Construction for Tunnel Mode...........31
5.1.2.2 IPv6 -- Header Construction for Tunnel Mode...........32
5.2 Processing Inbound IP Traffic...................................33
5.2.1 Selecting and Using an SA or SA Bundle.....................33
5.2.2 Handling of AH and ESP tunnels.............................34
6. ICMP Processing (relevant to IPsec)................................35
6.1 PMTU/DF Processing..............................................36
6.1.1 DF Bit.....................................................36
6.1.2 Path MTU Discovery (PMTU)..................................36
6.1.2.1 Propagation of PMTU...................................36
6.1.2.2 Calculation of PMTU...................................37
6.1.2.3 Granularity of PMTU Processing........................37
6.1.2.4 PMTU Aging............................................38
7. Auditing...........................................................39
8. Use in Systems Supporting Information Flow Security................39
8.1 Relationship Between Security Associations and Data Sensitivity.40
8.2 Sensitivity Consistency Checking................................40
8.3 Additional MLS Attributes for Security Association Databases....41
8.4 Additional Inbound Processing Steps for MLS Networking..........41
8.5 Additional Outbound Processing Steps for MLS Networking.........41
8.6 Additional MLS Processing for Security Gateways.................42
9. Performance Issues.................................................42
10. Conformance Requirements..........................................43
11. Security Considerations...........................................43
12. Differences from RFC 1825.........................................43
Acknowledgements......................................................44
Appendix A -- Glossary................................................45
Appendix B -- Analysis/Discussion of PMTU/DF/Fragmentation Issues.....48
B.1 DF bit..........................................................48
B.2 Fragmentation...................................................48
B.3 Path MTU Discovery..............................................52
B.3.1 Identifying the Originating Host(s)........................53
B.3.2 Calculation of PMTU........................................55
B.3.3 Granularity of Maintaining PMTU Data.......................56
B.3.4 Per Socket Maintenance of PMTU Data........................57
B.3.5 Delivery of PMTU Data to the Transport Layer...............57
B.3.6 Aging of PMTU Data.........................................57
Appendix C -- Sequence Space Window Code Example......................58
Appendix D -- Categorization of ICMP messages.........................60
References............................................................63
Disclaimer............................................................64
Author Information....................................................65
Full Copyright Statement..............................................66

1. Introduction

1.1 Summary of Contents of Document

This memo specifies the base architecture for IPsec compliant
systems. The goal of the architecture is to provide various security
services for traffic at the IP layer, in both the IPv4 and IPv6
environments. This document describes the goals of such systems,
their components and how they fit together with each other and into
the IP environment. It also describes the security services offered
by the IPsec protocols, and how these services can be employed in the
IP environment. This document does not address all aspects of IPsec
architecture. Subsequent documents will address additional
architectural details of a more advanced nature, e.g., use of IPsec
in NAT environments and more complete support for IP multicast. The
following fundamental components of the IPsec security architecture
are discussed in terms of their underlying, required functionality.
Additional RFCs (see Section 1.3 for pointers to other documents)
define the protocols in (a), (c), and (d).

a. Security Protocols -- Authentication Header (AH) and
Encapsulating Security Payload (ESP)
b. Security Associations -- what they are and how they work,
how they are managed, associated proces sing
c. Key Management -- manual and automatic (The Internet Key
Exchange (IKE))
d. Algorithms for authentication and encryption

This document is not an overall Security Architecture for the
Internet; it addresses security only at the IP layer, provided
through the use of a combination of cryptographic and protocol
security mechanisms.

The keywords MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT, SHOULD,
SHOULD NOT, RECOMMENDED, MAY, and OPTIONAL, when they appear in this
document, are to be interpreted as described in RFC 2119 [Bra97].

1.2 Audience

The target audience for this document includes implementers of this
IP security technology and others interested in gaining a general
background understanding of this system. In particular, prospective
users of this technology (end users or system administrators) are
part of the target audience. A glossary is provided as an appendix

to help fill in gaps in background/vocabulary. This document assumes
that the reader is familiar with the Internet Protocol, related
networking technology, and general security terms and concepts.

1.3 Related Documents

As mentioned above, other documents provide detailed definitions of
some of the components of IPsec and of their inter-relationship.
They include RFCs on the following topics:

a. "IP Security Document Roadmap" [TDG97] -- a document
providing guidelines for specifications describing encryption
and authentication algorithms used in this system.
b. security protocols -- RFCs describing the Authentication
Header (AH) [KA98a] and Encapsulating Security Payload (ESP)
[KA98b] protocols.
c. algorithms for authentication and encryption -- a separate
RFC for each algorithm.
d. automatic key management -- RFCs on "The Internet Key
Exchange (IKE)" [HC98], "Internet Security Association and
Key Management Protocol (ISAKMP)" [MSST97],"The OAKLEY Key
Determination Protocol" [Orm97], and "The Internet IP
Security Domain of Interpretation for ISAKMP" [Pip98].

2. Design Objectives

2.1 Goals/Objectives/Requirements/Problem Description

IPsec is designed to provide interoperable, high quality,
cryptographically-based security for IPv4 and IPv6. The set of
security services offered includes access control, connectionless
integrity, data origin authentication, protection against replays (a
form of partial sequence integrity), confidentiality (encryption),
and limited traffic flow confidentiality. These services are
provided at the IP layer, offering protection for IP and/or upper
layer protocols.

These objectives are met through the use of two traffic security
protocols, the Authentication Header (AH) and the Encapsulating
Security Payload (ESP), and through the use of cryptographic key
management procedures and protocols. The set of IPsec protocols
employed in any context, and the ways in which they are employed,
will be determined by the security and system requirements of users,
applications, and/or sites/organizations.

When these mechanisms are correctly implemented and deployed, they
ought not to adversely affect users, hosts, and other Internet
components that do not employ these security mechanisms for

protection of their traffic. These mechanisms also are designed to
be algorithm-independent. This modularity permits selection of
different sets of algorithms without affecting the other parts of the
implementation. For example, different user communities may select
different sets of algorithms (creating cliques) if required.

A standard set of default algorithms is specified to facilitate
interoperability in the global Internet. The use of these
algorithms, in conjunction with IPsec traffic protection and key
management protocols, is intended to permit system and application
developers to deploy high quality, Internet layer, cryptographic
security technology.

2.2 Caveats and Assumptions

The suite of IPsec protocols and associated default algorithms are
designed to provide high quality security for Internet traffic.
However, the security offered by use of these protocols ultimately
depends on the quality of the their implementation, which is outside
the scope of this set of standards. Moreover, the security of a
computer system or network is a function of many factors, including
personnel, physical, procedural, compromising emanations, and
computer security practices. Thus IPsec is only one part of an
overall system security architecture.

Finally, the security afforded by the use of IPsec is critically
dependent on many aspects of the operating environment in which the
IPsec implementation executes. For example, defects in OS security,
poor quality of random number sources, sloppy system management
protocols and practices, etc. can all degrade the security provided
by IPsec. As above, none of these environmental attributes are
within the scope of this or other IPsec standards.

3. System Overview

This section provides a high level description of how IPsec works,
the components of the system, and how they fit together to provide
the security services noted above. The goal of this description is
to enable the reader to "picture" the overall process/system, see how
it fits into the IP environment, and to provide context for later
sections of this document, which describe each of the components in
more detail.

An IPsec implementation operates in a host or a security gateway
environment, affording protection to IP traffic. The protection
offered is based on requirements defined by a Security Policy
Database (SPD) established and maintained by a user or system
administrator, or by an application operating within constraints

established by either of the above. In general, packets are selected
for one of three processing modes based on IP and transport layer
header information (Selectors, Section 4.4.2) matched against entries
in the database (SPD). Each packet is either afforded IPsec security
services, discarded, or allowed to bypass IPsec, based on the
applicable database policies identified by the Selectors.

3.1 What IPsec Does

IPsec provides security services at the IP layer by enabling a system
to select required security protocols, determine the algorithm(s) to
use for the service(s), and put in place any cryptographic keys
required to provide the requested services. IPsec can be used to
protect one or more "paths" between a pair of hosts, between a pair
of security gateways, or between a security gateway and a host. (The
term "security gateway" is used throughout the IPsec documents to
refer to an intermediate system that implements IPsec protocols. For
example, a router or a firewall implementing IPsec is a security
gateway.)

The set of security services that IPsec can provide includes access
control, connectionless integrity, data origin authentication,
rejection of replayed packets (a form of partial sequence integrity),
confidentiality (encryption), and limited traffic flow
confidentiality. Because these services are provided at the IP
layer, they can be used by any high er layer protocol, e.g., TCP, UDP,
ICMP, BGP, etc.

The IPsec DOI also supports negotiation of IP compression [SMPT98],
motivated in part by the observation that when encryption is employed
within IPsec, it prevents effective compression by lower protocol
layers.

3.2 How IPsec Works

IPsec uses two protocols to provide traffic security --
Authentication Header (AH) and Encapsulating Security Payload (ESP).
Both protocols are described in more detail in their respective RFCs
[KA98a, KA98b].

o The IP Authentication Header (AH) [KA98a] provides
connectionless integrity, data origin authentication, and an
optional anti-replay service.
o The Encapsulating Security Payload (ESP) protocol [KA98b] may
provide confidentiality (encryption), and limited traffic flow
confidentiality. It also may provide connectionless

integrity, data origin authentication, and an anti-replay
service. (One or the other set of these security services
must be applied whenever ESP is invoked.)
o Both AH and ESP are vehicles for access control, based on the
distribution of cryptographic keys and the management of
traffic flows relative to these security protocols.

These protocols may be applied alone or in combination with each
other to provide a desired set of security services in IPv4 and IPv6.
Each protocol supports two modes of use: transport mode and tunnel
mode. In transport mode the protocols provide protection primarily
for upper layer protocols; in tunnel mode, the protocols are applied
to tunneled IP packets. The differences between the two modes are
discussed in Section 4.

IPsec allows the user (or system administrator) to control the
granularity at which a security service is offered. For example, one
can create a single encrypted tunnel to carry all the traffic between
two security gateways or a separate encrypted tunnel can be created
for each TCP connection between each pair of hosts communicating
across these gateways. IPsec management must incorporate facilities
for specifying:

o which security services to use and in what combinations
o the granularity at which a given security protection should be
applied
o the algorithms used to effect cryptographic-based security

Because these security services use shared secret values
(cryptographic keys), IPsec relies on a separate set of mechanisms
for putting these keys in place. (The keys are used for
authentication/integrity and encryption services.) This document
requires support for both manual and automatic distribution of keys.
It specifies a specific public-key based approach (IKE -- [MSST97,
Orm97, HC98]) for automatic key management, but other automated key
distribution techniques MAY be used. For example, KDC-based systems
such as Kerberos and other public-key systems such as SKIP could be
employed.

3.3 Where IPsec May Be Implemented

There are several ways in which IPsec may be implemented in a host or
in conjunction with a router or firewall (to create a security
gateway). Several common examples are provided below:

a. Integration of IPsec into the native IP implementation. This
requires access to the IP source code and is applicable to
both hosts and security gateways.

b. "Bump-in-the-stack" (BITS) implementations, where IPsec is
implemented "underneath" an existing implementation of an IP
protocol stack, between the native IP and the local network
drivers. Source code access for the IP stack is not required
in this context, making this implementation approach
appropriate for use with legacy systems. This approach, when
it is adopted, is usually employed in hosts.

c. The use of an outboard crypto processor is a common design
feature of network security systems used by the military, and
of some commercial systems as well. It is sometimes referred
to as a "Bump-in-the-wire" (BITW) implementation. Such
implementations may be designed to serve either a host or a
gateway (or both). Usually the BITW device is IP
addressable. When supporting a single host, it may be quite
analogous to a BITS implementation, but in supporting a
router or firewall, it must operate like a security gateway.

4. Security Associations

This section defines Security Association management requirements for
all IPv6 implementations and for those IPv4 implementations that
implement AH, ESP, or both. The concept of a "Security Association"
(SA) is fundamental to IPsec. Both AH and ESP make use of SAs and a
major function of IKE is the establishment and maintenance of
Security Associations. All implementations of AH or ESP MUST support
the concept of a Security Association as described below. The
remainder of this section describes various aspects of Security
Association management, defining required characteristics for SA
policy management, traffic processing, and SA management techniques.

4.1 Definition and Scope

A Security Association (SA) is a simplex "connection" that affords
security services to the traffic carried by it. Security services
are afforded to an SA by the use of AH, or ESP, but not both. If
both AH and ESP protection is applied to a traffic stream, then two
(or more) SAs are created to afford protection to the traffic stream.
To secure typical, bi-directional communication between two hosts, or
between two security gateways, two Security Associations (one in each
direction) are required.

A security association is uniquely identified by a triple consisting
of a Security Parameter Index (SPI), an IP Destination Address, and a
security protocol (AH or ESP) identifier. In principle, the
Destination Address may be a unicast address, an IP broadcast
address, or a multicast group address. However, IPsec SA management
mechanisms currently are defined only for unicast SAs. Hence, in the

discussions that follow, SAs will be described in the context of
point-to-point communication, even though the concept is applicable
in the point-to-multipoint case as well.

As noted above, two types of SAs are defined: transport mode and
tunnel mode. A transport mode SA is a security association between
two hosts. In IPv4, a transport mode security protocol header
appears immediately after the IP header and any options, and before
any higher layer protocols (e.g., TCP or UDP). In IPv6, the security
protocol header appears after the base IP header and extensions, but
may appear before or after destination options, and before higher
layer protocols. In the case of ESP, a transport mode SA provides
security services only for these higher layer protocols, not for the
IP header or any extension headers preceding the ESP header. In the
case of AH, the protection is also extended to selected portions of
the IP header, selected portions of extension headers, and selected
options (contained in the IPv4 header, IPv6 Hop-by-Hop extension
header, or IPv6 Destination extension headers). For more details on
the coverage afforded by AH, see the AH specification [ KA98a].

A tunnel mode SA is essentially an SA applied to an IP tunnel.
Whenever either end of a security association is a security gateway,
the SA MUST be tunnel mode. Thus an SA between two security gateways
is always a tunnel mode SA, as is an SA between a host and a security
gateway. Note that for the case where traffic is destined for a
security gateway, e.g., SNMP commands, the security gateway is acting
as a host and transport mode is allowed. But in that case, the
security gateway is not acting as a gateway, i.e., not transiting
traffic. Two hosts MAY establish a tunnel mode SA between
themselves. The requirement for any (transit traffic) SA involving a
security gateway to be a tunnel SA arises due to the need to avoid
potential problems with regard to fragmentation and reassembly of
IPsec packets, and in circumstances where multiple paths (e.g., via
different security gateways) exist to the same destination behind the
security gateways.

For a tunnel mode SA, there is an "outer" IP header that specifies
the IPsec processing destination, plus an "inner" IP header that
specifies the (apparently) ultimate destination for the packet. The
security protocol header appears after the outer IP header, and
before the inner IP header. If AH is employed in tunnel mode,
portions of the outer IP header are afforded protection (as above),
as well as all of the tunneled IP packet (i.e., all of the inner IP
header is protected, as well as higher layer protocols). If ESP is
employed, the protection is afforded only to the tunneled packet, not
to the outer header.

In summary,
a) A host MUST support both transport and tunnel mode.
b) A security gateway is required to support only tunnel
mode. If it supports transport mode, that should be used
only when the security gateway is acting as a host, e.g.,
for network management.

4.2 Security Association Functionality

The set of security services offered by an SA depends on the security
protocol selected, the SA mode, the endpoints of the SA, and on the
election of optional services within the protocol. For example, AH
provides data origin authentication and connectionless integrity for
IP datagrams (hereafter referred to as just "authentication"). The
"precision" of the authentication service is a function of the
granularity of the security association with which AH is employed, as
discussed in Section 4.4.2, "Selectors".

AH also offers an anti-replay (partial sequence integrity) service at
the discretion of the receiver, to help counter denial of service
attacks. AH is an appropriate protocol to employ when
confidentiality is not required (or is not permitted, e.g , due to
government restrictions on use of encryption). AH also provides
authentication for selected portions of the IP header, which may be
necessary in some contexts. For example, if the integrity of an IPv4
option or IPv6 extension header must be protected en route between
sender and receiver, AH can provide this service (except for the
non-predictable but mutable parts of the IP header.)

ESP optionally provides confidentiality for traffic. (The strength
of the confidentiality service depends in part, on the encryption
algorithm employed.) ESP also may optionally provide authentication
(as defined above). If authentication is negotiated for an ESP SA,
the receiver also may elect to enforce an anti-replay service with
the same features as the AH anti-replay service. The scope of the
authentication offered by ESP is narrower than for AH, i.e., the IP
header(s) "outside" the ESP header is(are) not protected. If only
the upper layer protocols need to be authenticated, then ESP
authentication is an appropriate choice and is more space efficient
than use of AH encapsulating ESP. Note that although both
confidentiality and authentication are optional, they cannot both be
omitted. At least one of them MUST be selected.

If confidentiality service is selected, then an ESP (tunnel mode) SA
between two security gateways can offer partial traffic flow
confidentiality. The use of tunnel mode allows the inner IP headers
to be encrypted, concealing the identities of the (ultimate) traffic
source and destination. Moreover, ESP payload padding also can be

invoked to hide the size of the packets, further concealing the
external characteristics of the traffic. Similar traffic flow
confidentiality services may be offered when a mobile user is
assigned a dynamic IP address in a dialup context, and establishes a
(tunnel mode) ESP SA to a corporate firewall (acting as a security
gateway). Note that fine granularity SAs generally are more
vulnerable to traffic analysis than coarse granularity ones which are
carrying traffic from many subscribers.

4.3 Combining Security Associations

The IP datagrams transmitted over an individual SA are afforded
protection by exactly one security protocol, either AH or ESP, but
not both. Sometimes a security policy may call for a combination of
services for a particular traffic flow that is not achievable with a
single SA. In such instances it will be necessary to employ multiple
SAs to implement the required security policy. The term "security
association bundle" or "SA bundle" is applied to a sequence of SAs
through which traffic must be processed to satisfy a security policy.
The order of the sequence is defined by the policy. (Note that the
SAs that comprise a bundle may terminate at different endpoints. For
example, one SA may extend between a mobile host and a security
gateway and a second, nested SA may extend to a host behind the
gateway.)

Security associations may be combined into bundles in two ways:
transport adjacency and iterated tunneling.

o Transport adjacency refers to applying more than one
security protocol to the same IP datagram, without invoking
tunneling. This approach to combining AH and ESP allows
for only one level of combination; further nesting yields
no added benefit (assuming use of adequately strong
algorithms in each protocol) since the processing is
performed at one IPsec instance at the (ultimate)
destination.

Host 1 --- Security ---- Internet -- Security --- Host 2
| | Gwy 1 Gwy 2 | |
| | | |
| -----Security Association 1 (ESP transport)------- |
| |
-------Security Association 2 (AH transport)----------

o Iterated tunneling refers to the application of multiple
layers of security protocols effected through IP tunneling.
This approach allows for multiple levels of nesting, since
each tunnel can originate or terminate at a different IPsec

site along the path. No special treatment is expected for
ISAKMP traffic at intermediate security gateways other than
what can be specified through appropriate SPD entries (See
Case 3 in Section 4.5)

There are 3 basic cases of iterated tunneling -- support is
required only for cases 2 and 3.:

1. both endpoints for the SAs are the same -- The inner and
outer tunnels could each be either AH or ESP, though it
is unlikely that Host 1 would specify both to be the
same, i.e., AH inside of AH or ESP inside of ESP.

Host 1 --- Security ---- Internet -- Security --- Host 2
| | Gwy 1 Gwy 2 | |
| | | |
| -------Security Association 1 (tunnel)---------- | |
| |
---------Security Association 2 (tunnel)--------------

2. one endpoint of the SAs is the same -- The inner and
uter tunnels could each be either AH or ESP.

Host 1 --- Security ---- Internet -- Security --- Host 2
| | Gwy 1 Gwy 2 |
| | | |
| ----Security Association 1 (tunnel)---- |
| |
---------Security Association 2 (tunnel)-------------

3. neither endpoint is the same -- The inner and outer
tunnels could each be either AH or ESP.

Host 1 --- Security ---- Internet -- Security --- Host 2
| Gwy 1 Gwy 2 |
| | | |
| --Security Assoc 1 (tunnel)- |
| |
-----------Security Association 2 (tunnel)-----------

These two approaches also can be combined, e.g., an SA bundle could
be constructed from one tunnel mode SA and one or two transport mode
SAs, applied in sequence. (See Section 4.5 "Basic Combinations of
Security Associations.") Note that nested tunnels can also occur
where neither the source nor the destination endpoints of any of the
tunnels are the same. In that case, there would be no host or
security gateway with a bundle corresponding to the nested tunnels.

For transport mode SAs, only one ordering of security protocols seems
appropriate. AH is applied to both the upper layer protocols and
(parts of) the IP header. Thus if AH is used in a transport mode, in
conjunction with ESP, AH SHOULD appear as the first header after IP,
prior to the appearance of ESP. In that context, AH is applied to
the ciphertext output of ESP. In contrast, for tunnel mode SAs, one
can imagine uses for various orderings of AH and ESP. The required
set of SA bundle types that MUST be supported by a compliant IPsec
implementation is described in Section 4.5.

4.4 Security Association Databases

Many of the details associated with processing IP traffic in an IPsec
implementation are largely a local matter, not subject to
standardization. However, some external aspects of the processing
must be standardized, to ensure interoperability and to provide a
minimum management capability that is essential for productive use of
IPsec. This section describes a general model for processing IP
traffic relative to security associations, in support of these
interoperability and functionality goals. The model described below
is nominal; compliant implementations need not match details of this
model as presented, but the external behavior of such implementations
must be mappable to the externally observable characteristics of this
model.

There are two nominal databases in this model: the Security Policy
Database and the Security Association Database. The former specifies
the policies that determine the disposition of all IP traffic inbound
or outbound from a host, security gateway, or BITS or BITW IPsec
implementation. The latter database contains parameters that are
associated with each (active) security association. This section
also defines the concept of a Selector, a set of IP and upper layer
protocol field values that is used by the Security Policy Database to
map traffic to a policy, i.e., an SA (or SA bundle).

Each interface for which IPsec is enabled requires nominally separate
inbound vs. outbound databases (SAD and SPD), because of the
directionality of many of the fields that are used as selectors.
Typically there is just one such interface, for a host or security
gateway (SG). Note that an SG would always have at least 2
interfaces, but the "internal" one to the corporate net, usually
would not have IPsec enabled and so only one pair of SADs and one
pair of SPDs would be needed. On the other hand, if a host had
multiple interfaces or an SG had multiple external interfaces, it
might be necessary to have separate SAD and SPD pairs for each
interface.

4.4.1 The Security Policy Database (SPD)

Ultimately, a security association is a management construct used to
enforce a security policy in the IPsec environment. Thus an
essential element of SA processing is an underlying Security Policy
Database (SPD) that specifies what services are to be offered to IP
datagrams and in what fashion. The form of the database and its
interface are outside the scope of this specification. However, this
section does specify certain minimum management functionality that
must be provided, to allow a user or system administrator to control
how IPsec is applied to traffic transmitted or received by a host or
transiting a security gateway.

The SPD must be consulted during the processing of all traffic
(INBOUND and OUTBOUND), including non-IPsec traffic. In order to
support this, the SPD requires distinct entries for inbound and
outbound traffic. One can think of this as separate SPDs (inbound
vs. outbound). In addition, a nominally separate SPD must be
provided for each IPsec-enabled interface.

An SPD must discriminate among traffic that is afforded IPsec
protection and traffic that is allowed to bypass IPsec. This applies
to the IPsec protection to be applied by a sender and to the IPsec
protection that must be present at the receiver. For any outbound or
inbound datagram, three processing choices are possible: discard,
bypass IPsec, or apply IPsec. The first choice refers to traffic
that is not allowed to exit the host, traverse the security gateway,
or be delivered to an application at all. The second choice refers
to traffic that is allowed to pass without additional IPsec
protection. The third choice refers to traffic that is afforded
IPsec protection, and for such traffic the SPD must specify the
security services to be provided, protocols to be employed,
algorithms to be used, etc.

For every IPsec implementation, there MUST be an administrative
interface that allows a user or system administrator to manage the
SPD. Specifically, every inbound or outbound packet is subject to
processing by IPsec and the SPD must specify what action will be
taken in each case. Thus the administrative interface must allow the
user (or system administrator) to specify the security processing to
be applied to any packet entering or exiting the system, on a packet
by packet basis. (In a host IPsec implementation making use of a
socket interface, the SPD may not need to be consulted on a per
packet basis, but the effect is still the same.) The management
interface for the SPD MUST allow creation of entries consistent with
the selectors defined in Section 4.4.2, and MUST support (total)
ordering of these entries. It is expected that through the use of
wildcards in various selector fields, and because all packets on a

single UDP or TCP connection will tend to match a single SPD entry,
this requirement will not impose an unreasonably detailed level of
SPD specification. The selectors are analogous to what are found in
a stateless firewall or filtering router and which are currently
manageable this way.

In host systems, applications MAY be allowed to select what security
processing is to be applied to the traffic they generate and consume.
(Means of signalling such requests to the IPsec implementation are
outside the scope of this standard.) However, the system
administrator MUST be able to specify whether or not a user or
application can override (default) system policies. Note that
application specified policies may satisfy system requirements, so
that the system may not need to do additional IPsec processing beyond
that needed to meet an application's requirements. The form of the
management interface is not specified by this document and may differ
for hosts vs. security gateways, and within hosts the interface may
differ for socket-based vs. BITS implementations. However, this
document does specify a standard set of SPD elements that all IPsec
implementations MUST support.

The SPD contains an ordered list of policy entries. Each policy
entry is keyed by one or more selectors that define the set of IP
traffic encompassed by this policy entry. (The required selector
types are defined in Section 4.4.2.) These define the granularity of
policies or SAs. Each entry includes an indication of whether
traffic matching this policy will be bypassed, discarded, or subject
to IPsec processing. If IPsec processing is to be applied, the entry
includes an SA (or SA bundle) specification, listing the IPsec
protocols, modes, and algorithms to be employed, including any
nesting requirements. For example, an entry may call for all
matching traffic to be protected by ESP in transport mode using
3DES-CBC with an explicit IV, nested inside of AH in tunnel mode
using HMAC/SHA-1. For each selector, the policy entry specifies how
to derive the corresponding values for a new Security Association
Database (SAD, see Section 4.4.3) entry from those in the SPD and the
packet (Note that at present, ranges are only supported for IP
addresses; but wildcarding can be expressed for all selectors):

a. use the value in the packet itself -- This will limit use
of the SA to those packets which have this packet's value
for the selector even if the selector for the policy entry
has a range of allowed values or a wildcard for this
selector.
b. use the value associated with the policy entry -- If this
were to be just a single value, then there would be no
difference between (b) and (a). However, if the allowed
values for the selector are a range (for IP addresses) or

wildcard, then in the case of a range,(b) would enable use
of the SA by any packet with a selector value within the
range not just by packets with the selector value of the
packet that triggered the creation of the SA. In the case
of a wildcard, (b) would allow use of the SA by packets
with any value for this selector.

For example, suppose there is an SPD entry where the allowed value
for source address is any of a range of hosts (192.168.2.1 to
192.168.2.10). And suppose that a packet is to be sent that has a
source address of 192.168.2.3. The value to be used for the SA could
be any of the sample values below depending on what the policy entry
for this selector says is the source of the selector value:

source for the example of
value to be new SAD
used in the SA selector value
--------------- ------------
a. packet 192.168.2.3 (one host)
b. SPD entry 192.168.2.1 to 192.168.2.10 (range of hosts)

Note that if the SPD entry had an allowed value of wildcard for the
source address, then the SAD selector value could be wildcard (any
host). Case (a) can be used to prohibit sharing, even among packets
that match the same SPD entry.

As described below in Section 4.4.3, selectors may include "wildcard"
entries and hence the selectors for two entries may overlap. (This
is analogous to the overlap that arises with ACLs or filter entries
in routers or packet filtering firewalls.) Thus, to ensure
consistent, predictable processing, SPD entries MUST be ordered and
the SPD MUST always be searched in the same order, so that the first
matching entry is consistently selected. (This requirement is
necessary as the effect of processing traffic against SPD entries
must be deterministic, but there is no way to canonicalize SPD
entries given the use of wildcards for some selectors.) More detail
on matching of packets against SPD entries is provided in Section 5.

Note that if ESP is specified, either (but not both) authentication
or encryption can be omitted. So it MUST be possible to configure
the SPD value for the authentication or encryption algorithms to be
"NULL". However, at least one of these services MUST be selected,
i.e., it MUST NOT be possible to configure both of them as "NULL".

The SPD can be used to map traffic to specific SAs or SA bundles.
Thus it can function both as the reference database for security
policy and as the map to existing SAs (or SA bundles). (To
accommodate the bypass and discard policies cited above, the SPD also

MUST provide a means of mapping traffic to these functions, even
though they are not, per se, IPsec processing.) The way in which the
SPD operates is different for inbound vs. outbound traffic and it
also may differ for host vs. security gateway, BITS, and BITW
implementations. Sections 5.1 and 5.2 describe the use of the SPD
for outbound and inbound processing, respectively.

Because a security policy may require that more than one SA be
applied to a specified set of traffic, in a specific order, the
policy entry in the SPD must preserve these ordering requirements,
when present. Thus, it must be possible for an IPsec implementation
to determine that an outbound or inbound packet must be processed
thorough a sequence of SAs. Conceptually, for outbound processing,
one might imagine links (to the SAD) from an SPD entry for which
there are active SAs, and each entry would consist of either a single
SA or an ordered list of SAs that comprise an SA bundle. When a
packet is matched against an SPD entry and there is an existing SA or
SA bundle that can be used to carry the traffic, the processing of
the packet is controlled by the SA or SA bundle entry on the list.
For an inbound IPsec packet for which mul tiple IPsec SAs are to be
applied, the lookup based on destination address, IPsec protocol, and
SPI should identify a single SA.

The SPD is used to control the flow of ALL traffic through an IPsec
system, including security and key management traffic (e.g., ISAKMP)
from/to entities behind a security gateway. This means that ISAKMP
traffic must be explicitly accounted for in the SPD, else it will be
discarded. Note that a security gateway could prohibit traversal of
encrypted packets in various ways, e.g., having a DISCARD entry in
the SPD for ESP packets or providing proxy key exchange. In the
latter case, the traffic would be internally routed to the key
management module in the security gateway.

4.4.2 Selectors

An SA (or SA bundle) may be fine-grained or coarse-grained, depending
on the selectors used to define the set of traffic for the SA. For
example, all traffic between two hosts may be carried via a single
SA, and afforded a uniform set of security services. Alternatively,
traffic between a pair of hosts might be spread over multiple SAs,
depending on the applications being used (as defined by the Next
Protocol and Port fields), with different security services offered
by different SAs. Similarly, all traffic between a pair of security
gateways could be carried on a single SA, or one SA could be assigned
for each communicating host pair. The following selector parameters
MUST be supported for SA management to facilitate control of SA
granularity. Note that in the case of receipt of a packet with an
ESP header, e.g., at an encapsulating security gateway or BITW

implementation, the transport layer protocol, source/destination
ports, and Name (if present) may be "OPAQUE", i.e., inaccessible
because of encryption or fragmentation. Note also that both Source
and Destination addresses should either be IPv4 or IPv6.

- Destination IP Address (IPv4 or IPv6): this may be a single IP
address (unicast, anycast, broadcast (IPv4 only), or multicast
group), a range of addresses (high and low values (inclusive),
address + mask, or a wildcard address. The last three are used
to support more than one destination system sharing the same SA
(e.g., behind a security gateway). Note that this selector is
conceptually different from the "Destination IP Address" field
in the <Destination IP Address, IPsec Protocol, SPI> tuple used
to uniquely identify an SA. When a tunneled packet arrives at
the tunnel endpoint, its SPI/Destination address/Protocol are
used to look up the SA for this packet in the SAD. This
destination address comes from the encapsulating IP header.
Once the packet has been processed according to the tunnel SA
and has come out of the tunnel, its selectors are "looked up" in
the Inbound SPD. The Inbound SPD has a selector called
destination address. This IP destination address is the one in
the inner (encapsulated) IP header. In the case of a
transport'd packet, there will be only one IP header and this
ambiguity does not exist. [REQUIRED for all implementations]

- Source IP Address(es) (IPv4 or IPv6): this may be a single IP
address (unicast, anycast, broadcast (IPv4 only), or multicast
group), range of addresses (high and low values inclusive),
address + mask, or a wildcard address. The last three are used
to support more than one source system sharing the same SA
(e.g., behind a security gateway or in a multihomed host).
[REQUIRED for all implementations]

- Name: There are 2 cases (Note that these name forms are
supported in the IPsec DOI.)
1. User ID
a. a fully qualified user name string (DNS), e.g.,
mozart@foo.bar.com
b. X.500 distinguished name, e.g., C = US, SP = MA,
O = GTE Internetworking, CN = Stephen T. Kent.
2. System name (host, security gateway, etc.)
a. a fully qualified DNS name, e.g., foo.bar.com
b. X.500 distinguished name
c. X.500 general name

NOTE: One of the possible values of this selector is "OPAQUE".

[REQUIRED for the following cases. Note that support for name
forms other than addresses is not required for manually keyed
SAs.
o User ID
- native host implementations
- BITW and BITS implementations acting as HOSTS
with only one user
- security gateway implementations for INBOUND
processing.
o System names -- all implementations]

- Data sensitivity level: (IPSO/CIPSO labels)
[REQUIRED for all systems providing information flow security as
per Section 8, OPTIONAL for all other systems.]

- Transport Layer Protocol: Obtained from the IPv4 "Protocol" or
the IPv6 "Next Header" fields. This may be an individual
protocol number. These packet fields may not contain the
Transport Protocol due to the presence of IP extension headers,
e.g., a Routing Header, AH, ESP, Fragmentation Header,
Destination Options, Hop-by-hop options, etc. Note that the
Transport Protocol may not be available in the case of receipt
of a packet with an ESP header, thus a value of "OPAQUE" SHOULD
be supported.
[REQUIRED for all implementations]

NOTE: To locate the transport protocol, a system has to chain
through the packet headers checking the "Protocol" or "Next
Header" field until it encounters either one it recognizes as a
transport protocol, or until it reaches one that isn't on its
list of extension headers, or until it encounters an ESP header
that renders the transport protocol opaque.

- Source and Destination (e.g., TCP/UDP) Ports: These may be
individual UDP or TCP port values or a wildcard port. (The use
of the Next Protocol field and the Source and/or Destination
Port fields (in conjunction with the Source and/or Destination
Address fields), as an SA selector is sometimes referred to as
"session-oriented keying."). Note that the source and
destination ports may not be available in the case of receipt of
a packet with an ESP header, thus a value of "OPAQUE" SHOULD be
supported.

The following table summarizes the relationship between the
"Next Header" value in the packet and SPD and the derived Port
Selector value for the SPD and SAD.

Next Hdr Transport Layer Derived Port Selector Field
in Packet Protocol in SPD Value in SPD and SAD
-------- --------------- ---------------------------
ESP ESP or ANY ANY (i.e., don't look at it)
-don't care- ANY ANY (i.e., don't look at it)
specific value specific value NOT ANY (i.e., drop packet)
fragment
specific value specific value actual port selector field
not fragment

If the packet has been fragmented, then the port information may
not be avail able in the current fragment. If so, discard the
fragment. An ICMP PMTU should be sent for the first fragment,
which will have the port information. [MAY be supported]

The IPsec implementation context determines how selectors are used.
For example, a host implementation integrated into the stack may make
use of a socket interface. When a new connection is established the
SPD can be consulted and an SA (or SA bundle) bound to the socket.
Thus traffic sent via that socket need not result in additional
lookups to the SPD/SAD. In contrast, a BITS, BITW, or security
gateway implementation needs to look at each packet and perform an
SPD/SAD lookup based on the selectors. The allowable values for the
selector fields differ between the traffic flow, the security
association, and the security policy.

The following table summarizes the kinds of entries that one needs to
be able to express in the SPD and SAD. It shows how they relate to
the fields in data traffic being subjected to IPsec screening.
(Note: the "wild" or "wildcard" entry for src and dst addresses
includes a mask, range, etc.)

Field Traffic Value SAD Entry SPD Entry
-------- ------------- ---------------- --------------------
src addr single IP addr single,range,wild single,range,wildcard
dst addr single IP addr single,range,wild single,range,wildcard
xpt protocol* xpt protocol single,wildcard single,wildcard
src port* single src port single,wildcard single,wildcard
dst port* single dst port single,wildcard single,wildcard
user id* single user id single,wildcard single,wildcard
sec. labels single value single,wildcard single,wildcard

* The SAD and SPD entries for these fields could be "OPAQUE"
because the traffic value is encrypted.

NOTE: In principle, one could have selectors and/or selector values
in the SPD which cannot be negotiated for an SA or SA bundle.
Examples might include selector values used to select traffic for

discarding or enumerated lists which cause a separate SA to be
created for each item on the list. For now, this is left for future
versions of this document and the list of required selectors and
selector values is the same for the SPD and the SAD. However, it is
acceptable to have an administrative interface that supports use of
selector values which cannot be negotiated provided that it does not
mislead the user into believing it is creating an SA with these
selector values. For example, the interface may allow the user to
specify an enumerated list of values but would result in the creation
of a separate policy and SA for each item on the list. A vendor
might support such an interface to make it easier for its customers
to specify clear and concise policy specifications.

4.4.3 Security Association Database (SAD)

In each IPsec implementation there is a nominal Security Association
Database, in which each entry defines the parameters associated with
one SA. Each SA has an entry in the SAD. For outbound processing,
entries are pointed to by entries in the SPD. Note that if an SPD
entry does not currently point to an SA that is appropriate for the
packet, the implementation creates an appropriate SA (or SA Bundle)
and links the SPD entry to the SAD entry (see Section 5.1.1). For
inbound processing, each entry in the SAD is indexed by a destination
IP address, IPsec protocol type, and SPI. The following parameters
are associated with each entry in the SAD. This description does not
purport to be a MIB, but only a specification of the minimal data
items required to support an SA in an IPsec implementation.

For inbound processing: The following packet fields are used to look
up the SA in the SAD:

o Outer Header's Destination IP address: the IPv4 or IPv6
Destination address.
[REQUIRED for all implementations]
o IPsec Protocol: AH or ESP, used as an index for SA lookup
in this database. Specifies the IPsec protocol to be
applied to the traffic on this SA.
[REQUIRED for all implementations]
o SPI: the 32-bit value used to distinguish among different
SAs terminating at the same destination and using the same
IPsec protocol.
[REQUIRED for all implementations]

For each of the selectors defined in Section 4.4.2, the SA entry in
the SAD MUST contain the value or values which were negotiated at the
time the SA was created. For the sender, these values are used to
decide whether a given SA is appropriate for use with an outbound
packet. This is part of checking to see if there is an existing SA

that can be used. For the receiver, these values are used to check
that the selector values in an inbound packet match those for the SA
(and thus indirectly those for the matching policy). For the
receiver, this is part of verifying that the SA was appropriate for
this packet. (See Section 6 for rules for ICMP messages.) These
fields can have the form of specific values, ranges, wildcards, or
"OPAQUE" as described in section 4.4.2, "Selectors". Note that for
an ESP SA, the encryption algorithm or the authentication algorithm
could be "NULL". However they MUST not both be "NULL".

The following SAD fields are used in doing IPsec processing:

o Sequence Number Counter: a 32-bit value used to generate the
Sequence Number field in AH or ESP headers.
[REQUIRED for all implementations, but used only for outbound
traffic.]
o Sequence Counter Overflow: a flag indicating whether overflow
of the Sequence Number Counter should generate an auditable
event and prevent transmission of additional packets on the
SA.
[REQUIRED for all implementations, but used only for outbound
traffic.]
o Anti-Replay Window: a 32-bit counter and a bit-map (or
equivalent) used to determine whether an inbound AH or ESP
packet is a replay.
[REQUIRED for all implementations but used only for inbound
traffic. NOTE: If anti-replay has been disabled by the
receiver, e.g., in the case of a manually keyed SA, then the
Anti-Replay Window is not used.]
o AH Authentication algorithm, keys, etc.
[REQUIRED for AH implementations]
o ESP Encryption algorithm, keys, IV mode, IV, etc.
[REQUIRED for ESP implementations]
o ESP authentication algorithm, keys, etc. If the
authentication service is not selected, this field will be
null.
[REQUIRED for ESP implementations]
o Lifetime of this Security Association: a time interval after
which an SA must be replaced with a new SA (and new SPI) or
terminated, plus an indication of which of these actions
should occur. This may be expressed as a time or byte count,
or a simultaneous use of both, the first lifetime to expire
taking precedence. A compliant implementation MUST support
both types of lifetimes, and must support a simultaneous use
of both. If time is employed, and if IKE employs X.509
certificates for SA establishment, the SA lifetime must be
constrained by the validity intervals of the certificates,
and the NextIssueDate of the CRLs used in the IKE exchange

for the SA. Both initiator and responder are responsible for
constraining SA lifetime in this fashion.
[REQUIRED for all implementations]

NOTE: The details of how to handle the refreshing of keys
when SAs expire is a local matter. However, one reasonable
approach is:
(a) If byte count is used, then the implementation
SHOULD count the number of bytes to which the IPsec
algorithm is applied. For ESP, this is the encryption
algorithm (including Null encryption) and for AH,
this is the authentication algorithm. This includes
pad bytes, etc. Note that implementations SHOULD be
able to handle having the counters at the ends of an
SA get out of synch, e.g., because of packet loss or
because the implementations at each end of the SA
aren't doing things the same way.
(b) There SHOULD be two kinds of lifetime -- a soft
lifetime which warns the implementation to initiate
action such as setting up a replacement SA and a
hard lifetime when the current SA ends.
(c) If the entire packet does not get delivered during
the SAs lifetime, the packet SHOULD be discarded.

o IPsec protocol mode: tunnel, transport or wildcard.
Indicates which mode of AH or ESP is applied to traffic on
this SA. Note that if this field is "wildcard" at the
sending end of the SA, then the application has to specify
the mode to the IPsec implementation. This use of wildcard
allows the same SA to be used for either tunnel or transport
mode traffic on a per packet basis, e.g., by different
sockets. The receiver does not need to know the mode in
order to properly process the packet's IPsec headers.

[REQUIRED as follows, unless implicitly defined by context:
- host implementations must support all modes
- gateway implementations must support tunnel mode]

NOTE: The use of wildcard for the protocol mode of an inbound
SA may add complexity to the situation in the receiver (host
only). Since the packets on such an SA could be delivered in
either tunnel or transport mode, the security of an incoming
packet could depend in part on which mode had been used to
deliver it. If, as a result, an application cared about the
SA mode of a given packet, then the application would need a
mechanism to obtain this mode information.

o Path MTU: any observed path MTU and aging variables. See
Section 6.1.2.4
[REQUIRED for all implementations but used only for outbound
traffic]

4.5 Basic Combinations of Security Associations

This section describes four examples of combinations of security
associations that MUST be supported by compliant IPsec hosts or
security gateways. Additional combinations of AH and/or ESP in
tunnel and/or transport modes MAY be supported at the discretion of
the implementor. Compliant implementations MUST be capable of
generating these four combinations and on receipt, of processing
them, but SHOULD be able to receive and process any combination. The
diagrams and text below describe the basic cases. The legend for the
diagrams is:

==== = one or more security associations (AH or ESP, transport
or tunnel)
---- = connectivity (or if so labelled, administrative boundary)
Hx = host x
SGx = security gateway x
X* = X supports IPsec

NOTE: The security associations below can be either AH or ESP. The
mode (tunnel vs transport) is determined by the nature of the
endpoints. For host-to-host SAs, the mode can be either transport or
tunnel.

Case 1. The case of providing end-to-end security between 2 hosts
across the Internet (or an Intranet).

====================================
| |
H1* ------ (Inter/Intranet) ------ H2*

Note that either transport or tunnel mode can be selected by the
hosts. So the headers in a packet between H1 and H2 could look
like any of the following:

Transport Tunnel
----------------- ---------------------
1. [IP1][AH][upper] 4. [IP2][AH][IP1][upper]
2. [IP1][ESP][upper] 5. [IP2][ESP][IP1][upper]
3. [IP1][AH][ESP][upper]

Note that there is no requirement to support general nesting,
but in transport mode, both AH and ESP can be applied to the
packet. In this event, the SA establishment procedure MUST
ensure that first ESP, then AH are applied to the packet.

Case 2. This case illustrates simple virtual private networks
support.

===========================
| |
---------------------|---- ---|-----------------------
| | | | | |
| H1 -- (Local --- SG1* |--- (Internet) ---| SG2* --- (Local --- H2 |
| Intranet) | | Intranet) |
-------------------------- ---------------------------
admin. boundary admin. boundary

Only tunnel mode is required here. So the headers in a packet
between SG1 and SG2 could look like either of the following:

Tunnel
---------------------
4. [IP2][AH][IP1][upper]
5. [IP2][ESP][IP1][upper]

Case 3. This case combines cases 1 and 2, adding end-to-end security
between the sending and receiving hosts. It imposes no new
requirements on the hosts or security gateways, other than a
requirement for a security gateway to be configurable to pass
IPsec traffic (including ISAKMP traffic) for hosts behind it.

===============================================================
| |
| ========================= |
| | | |
---|-----------------|---- ---|-------------------|---
| | | | | | | |
| H1* -- (Local --- SG1* |-- (Internet) --| SG2* --- (Local --- H2* |
| Intranet) | | Intranet) |
-------------------------- ---------------------------
admin. boundary admin. boundary

Case 4. This covers the situation where a remote host (H1) uses the
Internet to reach an organization's firewall (SG2) and to then
gain access to some server or other machine (H2). The remote
host could be a mobile host (H1) dialing up to a local PPP/ARA
server (not shown) on the Internet and then crossing the
Internet to the home organization's firewall (SG2), etc. The

details of support for this case, (how H1 locates SG2,
authenticates it, and verifies its authorization to represent
H2) are discussed in Section 4.6.3, "Locating a Security
Gateway".

======================================================
| |
|============================== |
|| | |
|| ---|----------------------|---
|| | | | |
H1* ----- (Internet) ------| SG2* ---- (Local ----- H2* |
^ | Intranet) |
| ------------------------------
could be dialup admin. boundary (optional)
to PPP/ARA server

Only tunnel mode is required between H1 and SG2. So the choices
for the SA between H1 and SG2 would be one of the ones in case
2. The choices for the SA between H1 and H2 would be one of the
ones in case 1.

Note that in this case, the sender MUST apply the transport
header before the tunnel header. Therefore the management
interface to the IPsec implementation MUST support configuration
of the SPD and SAD to ensure this ordering of IPsec header
application.

As noted above, support for additional combinations of AH and ESP is
optional. Use of other, optional combinations may adversely affect
interoperability.

4.6 SA and Key Management

IPsec mandates support for both manual and automated SA and
cryptographic key management. The IPsec protocols, AH and ESP, are
largely independent of the associated SA management techniques,
although the techniques involved do affect some of the security
services offered by the protocols. For example, the optional anti-
replay services available for AH and ESP require automated SA
management. Moreover, the granularity of key distribution employed
with IPsec determines the granularity of authentication provided.
(See also a discussion of this issue in Section 4.7.) In general,
data origin authentication in AH and ESP is limited by the extent to
which secrets used with the authentication algorithm (or with a key
management protocol that creates such secrets) are shared among
multiple possible sources.

The following text describes the minimum requirements for both types
of SA management.

4.6.1 Manual Techniques

The simplest form of management is manual management, in which a
person manually configures each system with keying material and
security association management data relevant to secure communication
with other systems. Manual techniques are practical in small, static
environments but they do not scale well. For example, a company
could create a Virtual Private Network (VPN) using IPsec in security
gateways at several sites. If the number of sites is small, and
since all the sites come under the purview of a single administrative
domain, this is likely to be a feasible context for manual management
techniques. In this case, the security gateway might selectively
protect traffic to and from other sites within the organization using
a manually configured key, while not protecting traffic for other
destinations. It also might be appropriate when only selected
communications need to be secured. A similar argument might apply to
use of IPsec entirely within an organization for a small number of
hosts and/or gateways. Manual management techniques often employ
statically configured, symmetric keys, though other options also
exist.

4.6.2 Automated SA and Key Management

Widespread deployment and use of IPsec requires an Internet-standard,
scalable, automated, SA management protocol. Such support is
required to facilitate use of the anti-replay features of AH and ESP,
and to accommodate on-demand creation of SAs, e.g., for user- and
session-oriented keying. (Note that the notion of "rekeying" an SA
actually implies creation of a new SA with a new SPI, a process that
generally implies use of an automated SA/key management protocol.)

The default automated key management protocol selected for use with
IPsec is IKE [MSST97, Orm97, HC98] under the IPsec domain of
interpretation [Pip98]. Other automated SA management protocols MAY
be employed.

When an automated SA/key management protocol is employed, the output
from this protocol may be used to generate multiple keys, e.g., for a
single ESP SA. This may arise because:

o the encryption algorithm uses multiple keys (e.g., triple DES)
o the authentication algorithm uses multiple keys
o both encryption and authentication algorithms are employed

The Key Management System may provide a separate string of bits for
each key or it may generate one string of bits from which all of them
are extracted. If a single string of bits is provided, care needs to
be taken to ensure that the parts of the system that map the string
of bits to the required keys do so in the same fashion at both ends
of the SA. To ensure that the IPsec implementations at each end of
the SA use the same bits for the same keys, and irrespective of which
part of the system divides the string of bits into individual keys,
the encryption key(s) MUST be taken from the first (left-most, high-
order) bits and the authentication key(s) MUST be taken from the
remaining bits. The number of bits for each key is defined in the
relevant algorithm specification RFC. In the case of multiple
encryption keys or multiple authentication keys, the specification
for the algorithm must specify the order in which they are to be
selected from a single string of bits provided to the algorithm.

4.6.3 Locating a Security Gateway

This section discusses issues relating to how a host learns about the
existence of relevant security gateways and once a host has contacted
these security gateways, how it knows that these are the correct
security gateways. The details of where the required information is
stored is a local matter.

Consider a situation in which a remote host (H1) is using the
Internet to gain access to a server or other machine (H2) and there
is a security gateway (SG2), e.g., a firewall, through which H1's
traffic must pass. An example of this situation would be a mobile
host (Road Warrior) crossing the Internet to the home organization's
firewall (SG2). (See Case 4 in the section 4.5 Basic Combinations of
Security Associations.) This situation raises several issues:

1. How does H1 know/learn about the existence of the security
gateway SG2?
2. How does it authenticate SG2, and once it has authenticated
SG2, how does it confirm that SG2 has been authorized to
represent H2?
3. How does SG2 authenticate H1 and verify that H1 is authorized
to contact H2?
4. How does H1 know/learn about backup gateways which provide
alternate paths to H2?

To address these problems, a host or secu rity gateway MUST have an
administrative interface that allows the user/administrator to
configure the address of a security gateway for any sets of
destination addresses that require its use. This includes the ability
to configure:

o the requisite information for locating and authenticating the
security gateway and verifying its authorization to represent
the destination host.
o the requisite information for locating and authenticating any
backup gateways and verifying their authorization to represent
the destination host.

It is assumed that the SPD is also configured with policy information
that covers any other IPsec requirements for the path to the security
gateway and the destination host.

This document does not address the issue of how to automate the
discovery/verification of security gateways.

4.7 Security Associations and Multicast

The receiver-orientation of the Security Association implies that, in
the case of unicast traffic, the destination system will normally
select the SPI value. By having the destination select the SPI
value, there is no potential for manually configured Security
Associations to conflict with automatically configured (e.g., via a
key management protocol) Security Associations or for Security
Associations from multiple sources to conflict with each other. For
multicast traffic, there are multiple destination systems per
multicast group. So some system or person will need to coordinate
among all multicast groups to select an SPI or SPIs on behalf of each
multicast group and then communicate the group's IPsec information to
all of the legitimate members of that multicast group via mechanisms
not defined here.

Multiple senders to a multicast group SHOULD use a single Security
Association (and hence Security Parameter Index) for all traffic to
that group when a symmetric key encryption or authentication
algorithm is employed. In such circumstances, the receiver knows only
that the message came from a system possessing the key for that
multicast group. In such circumstances, a receiver generally will
not be able to authenticate which system sent the multicast traffic.
Specifications for other, more general multicast cases are deferred
to later IPsec documents.

At the time this specification was published, automated protocols for
multicast key distribution were not considered adequately mature for
standardization. For multicast groups having relatively few members,
manual key distribution or multiple use of existing unicast key
distribution algorithms such as modified Diffie-Hellman appears
feasible. For very large groups, new scalable techniques will be
needed. An example of current work in this area is the Group Key
Management Protocol (GKMP) [HM97].

5. IP Traffic Processing

As mentioned in Section 4.4.1 "The Security Policy Database (SPD)",
the SPD must be consulted during the processing of all traffic
(INBOUND and OUTBOUND), including non-IPsec traffic. If no policy is
found in the SPD that matches the packet (for either inbound or
outbound traffic), the packet MUST be discarded.

NOTE: All of the cryptographic algorithms used in IPsec expect their
input in canonical network byte order (see Appendix in RFC 791) and
generate their output in canonical network byte order. IP packets
are also transmitted in network byte order.

5.1 Outbound IP Traffic Processing

5.1.1 Selecting and Using an SA or SA Bundle

In a security gateway or BITW implementation (and in many BITS
implementations), each outbound packet is compared against the SPD to
determine what processing is required for the packet. If the packet
is to be discarded, this is an auditable event. If the traffic is
allowed to bypass IPsec processing, the packet continues through
"normal" processing for the environment in which the IPsec processing
is taking place. If IPsec processing is required, the packet is
either mapped to an existing SA (or SA bundle), or a new SA (or SA
bundle) is created for the packet. Since a packet's selectors might
match multiple policies or multiple extant SAs and since the SPD is
ordered, but the SAD is not, IPsec MUST:

1. Match the packet's selector fields against the outbound
policies in the SPD to locate the first appropriate
policy, which will point to zero or more SA bundles in the
SAD.

2. Match the packet's selector fields against those in the SA
bundles found in (1) to locate the first SA bundle that
matches. If no SAs were found or none match, create an
appropriate SA bundle and link the SPD entry to the SAD
entry. If no key management entity is found, drop the
packet.

3. Use the SA bundle found/created in (2) to do the required
IPsec processing, e.g., authenticate and encrypt.

In a host IPsec implementation based on sockets, the SPD will be
consulted whenever a new socket is created, to determine what, if
any, IPsec processing will be applied to the traffic that will flow
on that socket.

NOTE: A compliant implementation MUST not allow instantiation of an
ESP SA that employs both a NULL encryption and a NULL authentication
algorithm. An attempt to negotiate such an SA is an auditable event.

5.1.2 Header Construction for Tunnel Mode

This section describes the handling of the inner and outer IP
headers, extension headers, and options for AH and ESP tunnels. This
includes how to construct the encapsulating (outer) IP header, how to
handle fields in the inner IP header, and what other actions should
be taken. The general idea is modeled after the one used in RFC
2003, "IP Encapsulation with IP":

o The outer IP header Source Address and Destination Address
identify the "endpoints" of the tunnel (the encapsulator and
decapsulator). The inner IP header Source Address and
Destination Addresses identify the original sender and
recipient of the datagram, (from the perspective of this
tunnel), respectively. (see footnote 3 after the table in
5.1.2.1 for more details on the encapsulating source IP
address.)
o The inner IP header is not changed except to decrement the TTL
as noted below, and remains unchanged during its delivery to
the tunnel exit point.
o No change to IP options or extension headers in the inner
header occurs during delivery of the encapsulated datagram
through the tunnel.
o If need be, other protocol headers such as the IP
Authentication header may be inserted between the outer IP
header and the inner IP header.

The tables in the following sub-sections show the handling for the
different header/option fields (constructed = the value in the outer
field is constructed independently of the value in the inner).

5.1.2.1 IPv4 -- Header Construction for Tunnel Mode

<-- How Outer Hdr Relates to Inner Hdr -->
Outer Hdr at Inner Hdr at
IPv4 Encapsulator Decapsulator
Header fiel ds: -------------------- ------------
version 4 (1) no change
header length constructed no change
TOS copied from inner hdr (5) no change
total length constructed no change
ID constructed no change
flags (DF,MF) constructed, DF (4) no change
fragmt offset constructed no change

TTL constructed (2) decrement (2)
protocol AH, ESP, routing hdr no change
checksum constructed constructed (2)
src address constructed (3) no change
dest address constructed (3) no change
Options never copied no change

1. The IP version in the encapsulating header can be different
from the value in the inner header.

2. The TTL in the inner header is decremented by the
encapsulator prior to forwarding and by the decapsulator if
it forwards the packet. (The checksum changes when the TTL
changes.)

Note: The decrementing of the TTL is one of the usual actions
that takes place when forwarding a packet. Packets
originating from the same node as the encapsulator do not
have their TTL's decremented, as the sending node is
originating the packet rather than forwarding it.

3. src and dest addresses depend on the SA, which is used to
determine the dest address which in turn determines which src
address (net interface) is used to forward the packet.

NOTE: In principle, the encapsulating IP source address can
be any of the encapsulator's interface addresses or even an
address different from any of the encapsulator's IP
addresses, (e.g., if it's acting as a NAT box) so long as the
address is reachable through the encapsulator from the
environment into which the packet is sent. This does not
cause a problem because IPsec does not currently have any
INBOUND processing requirement that involves the Source
Address of the encapsulating IP header. So while the
receiving tunnel endpoint looks at the Destination Address in
the encapsulating IP header, it only looks at the Source
Address in the inner (encapsulated) IP header.

4. configuration determines whether to copy from the inner
header (IPv4 only), clear or set the DF.

5. If Inner Hdr is IPv4 (Protocol = 4), copy the TOS. If Inner
Hdr is IPv6 (Protocol = 41), map the Class to TOS.

5.1.2.2 IPv6 -- Header Construction for Tunnel Mode

See previous section 5.1.2 for notes 1-5 indicated by (footnote
number).

<-- How Outer Hdr Relates Inner Hdr --->
Outer Hdr at Inner Hdr at
IPv6 Encapsulator Decapsulator
Header fields: -------------------- ------------
version 6 (1) no change
class copied or configured (6) no change
flow id copied or configured no change
len constructed no change
next header AH,ESP,routing hdr no change
hop limit constructed (2) decrement (2)
src address constructed (3) no change
dest address constructed (3) no change
Extension headers never copied no change

6. If Inner Hdr is IPv6 (Next Header = 41), copy the Class. If
Inner Hdr is IPv4 (Next Header = 4), map the TOS to Class.

5.2 Processing Inbound IP Traffic

Prior to performing AH or ESP processing, any IP fragments are
reassembled. Each inbound IP datagram to which IPsec processing will
be applied is identified by the appearance of the AH or ESP values in
the IP Next Protocol field (or of AH or ESP as an extension header in
the IPv6 context).

Note: Appendix C contains sample code for a bitmask check for a 32
packet window that can be used for implementing anti-replay service.

5.2.1 Selecting and Using an SA or SA Bundle

Mapping the IP datagram to the appropriate SA is simplified because
of the presence of the SPI in the AH or ESP header. Note that the
selector checks are made on the inner headers not the outer (tunnel)
headers. The steps followed are:

1. Use the packet's destination address (outer IP header),
IPsec protocol, and SPI to look up the SA in the SAD. If
the SA lookup fails, drop the packet and log/report the
error.

2. Use the SA found in (1) to do the IPsec processing, e.g.,
authenticate and decrypt. This step includes matching the
packet's (Inner Header if tunneled) selectors to the
selectors in the SA. Local policy determines the
specificity of the SA selectors (single value, list,
range, wildcard). In general, a packet's source address
MUST match the SA selector value. However, an ICMP packet
received on a tunnel mode SA may have a source address

other than that bound to the SA and thus such packets
should be permitted as exceptions to this check. For an
ICMP packet, the selectors from the enclosed problem
packet (the source and destination addresses and ports
should be swapped) should be checked against the selectors
for the SA. Note that some or all of these selectors may
be inaccessible because of limitations on how many bits of
the problem packet the ICMP packet is allowed to carry or
due to encryption. See Section 6.

Do (1) and (2) for every IPsec header until a Transport
Protocol Header or an IP header that is NOT for this
system is encountered. Keep track of what SAs have been
used and their order of application.

3. Find an incoming policy in the SPD that matches the
packet. This could be done, for example, by use of
backpointers from the SAs to the SPD or by matching the
packet's selectors (Inner Header if tunneled) against
those of the policy entries in the SPD.

4. Check whether the required IPsec processing has been
applied, i.e., verify that the SA's found in (1) and (2)
match the kind and order of SAs required by the policy
found in (3).

NOTE: The correct "matching" policy will not necessarily
be the first inbound policy found. If the check in (4)
fails, steps (3) and (4) are repeated until all policy
entries have been checked or until the check succeeds.

At the end of these steps, pass the resulting packet to the Transport
Layer or forward the packet. Note that any IPsec headers processed
in these steps may have been removed, but that this information,
i.e., what SAs were used and th e order of their application, may be
needed for subsequent IPsec or firewall processing.

Note that in the case of a security gateway, if forwarding causes a
packet to exit via an IPsec-enabled interface, then additional IPsec
processing may be applied.

5.2.2 Handling of AH and ESP tunnels

The handling of the inner and outer IP headers, extension headers,
and options for AH and ESP tunnels should be performed as described
in the tables in Section 5.1.

6. ICMP Processing (relevant to IPsec)

The focus of this section is on the handling of ICMP error messages.
Other ICMP traffic, e.g., Echo/Reply, should be treated like other
traffic and can be protected on an end-to-end basis using SAs in the
usual fashion.

An ICMP error message protected by AH or ESP and generated by a
router SHOULD be processed and forwarded in a tunnel mode SA. Local
policy determines whether or not it is subjected to source address
checks by the router at the destination end of the tunnel. Note that
if the router at the originating end of the tunnel is forwarding an
ICMP error message from another router, the source address check
would fail. An ICMP message protected by AH or ESP and generated by
a router MUST NOT be forwarded on a transport mode SA (unless the SA
has been established to the router acting as a host, e.g., a Telnet
connection used to manage a router). An ICMP message generated by a
host SHOULD be checked against the source IP address selectors bound
to the SA in which the message arrives. Note that even if the source
of an ICMP error message is authenticated, the returned IP header
could be invalid. Accordingly, the selector values in the IP header
SHOULD also be checked to be sure that they are consistent with the
selectors for the SA over which the ICMP message was received.

The table in Appendix D characterize ICMP messages as being either
host generated, router generated, both, unknown/unassigned. ICMP
messages falling into the last two categories should be handled as
determined by the receiver's policy.

An ICMP message not protected by AH or ESP is unauthenticated and its
processing and/or forwarding may result in denial of service. This
suggests that, in general, it would be desirable to ignore such
messages. However, it is expected that many routers (vs. security
gateways) will not implement IPsec for transit traffic and thus
strict adherence to this rule would cause many ICMP messages to be
discarded. The result is that some critical IP functions would be
lost, e.g., redirection and PMTU processing. Thus it MUST be
possible to configure an IPsec implementation to accept or reject
(router) ICMP traffic as per local security policy.

The remainder of this section addresses how PMTU processing MUST be
performed at hosts and security gateways. It addresses processing of
both authenticated and unauthenticated ICMP PMTU messages. However,
as noted above, unauthenticated ICMP messages MAY be discarded based
on local policy.

6.1 PMTU/DF Processing

6.1.1 DF Bit

In cases where a system (host or gateway) adds an encapsulating
header (ESP tunnel or AH tunnel), it MUST support the option of
copying the DF bit from the original packet to the encapsulating
header (and processing ICMP PMTU messages). This means that it MUST
be possible to configure the system's treatment of the DF bit (set,
clear, copy from encapsulated header) for each interface. (See
Appendix B for rationale.)

6.1.2 Path MTU Discovery (PMTU)

This section discusses IPsec handling for Path MTU Discovery
messages. ICMP PMTU is used here to refer to an ICMP message for:

IPv4 (RFC 792):
- Type = 3 (Destination Unreachable)
- Code = 4 (Fragmentation needed and DF set)
- Next-Hop MTU in the low-order 16 bits of the second
word of the ICMP header (labelled "unused" in RFC
792), with high-order 16 bits set to zero

IPv6 (RFC 1885):
- Type = 2 (Packet Too Big)
- Code = 0 (Fragmentation needed)
- Next-Hop MTU in the 32 bit MTU field of the ICMP6
message

6.1.2.1 Propagation of PMTU

The amount of information returned with the ICMP PMTU message (IPv4
or IPv6) is limited and this affects what selectors are available for
use in further propagating the PMTU information. (See Appendix B for
more detailed discussion of this topic.)

o PMTU message with 64 bits of IPsec header -- If the ICMP PMTU
message contains only 64 bits of the IPsec header (minimum for
IPv4), then a security gateway MUST support the following options
on a per SPI/SA basis:

a. if the originating host can be determined (or the possible
sources narrowed down to a manageable number), send the PM
information to all the possible originating hosts.
b. if the originating host cannot be determined, store the PMTU
with the SA and wait until the next packet(s) arrive from the
originating host for the relevant security association. If

the packet(s) are bigger than the PMTU, drop the packet(s),
and compose ICMP PMTU message(s) with the new packet(s) and
the updated PMTU, and send the ICMP message(s) about the
problem to the originating host. Retain the PMTU information
for any message that might arrive subsequently (see Section
6.1.2.4, "PMTU Aging").

o PMTU message with >64 bits of IPsec header -- If the ICMP message
contains more information from the original packet then there may
be enough non-opaque information to immediately determine to which
host to propagate the ICMP/PMTU message and to provide that system
with the 5 fields (source address, destination address, source
port, destination port, transport protocol) needed to determine
where to store/update the PMTU. Under such circumstances, a
security gateway MUST generate an ICMP PMTU message immediately
upon receipt of an ICMP PMTU from further down the path.

o Distributing the PMTU to the Transport Layer -- The host mechanism
for getting the updated PMTU to the transport layer is unchanged,
as specified in RFC 1191 (Path MTU Discovery).

6.1.2.2 Calculation of PMTU

The calculation of PMTU from an ICMP PMTU MUST take into account the
addition of any IPsec header -- AH transport, ESP transport, AH/ESP
transport, ESP tunnel, AH tunnel. (See Appendix B for discussion of
implementation issues.)

Note: In some situations the addition of IPsec headers could result
in an effective PMTU (as seen by the host or application) that is
unacceptably small. To avoid this problem, the implementation may
establish a threshold below which it will not report a reduced PMTU.
In such cases, the implementation would apply IPsec and then fragment
the resulting packet according to the PMTU. This would result in a
more efficient use of the available bandwidth.

6.1.2.3 Granularity of PMTU Processing

In hosts, the granularity with which ICMP PMTU processing can be done
differs depending on the implementation situation. Looking at a
host, ther e are 3 situations that are of interest with respect to
PMTU issues (See Appendix B for additional details on this topic.):

a. Integration of IPsec into the native IP implementation
b. Bump-in-the-stack implementations, where IPsec is implemented
"underneath" an existing implementation of a TCP/IP protocol
stack, between the native IP and the local network drivers

c. No IPsec implementation -- This case is included because it
is relevant in cases where a security gateway is sending PMTU
information back to a host.

Only in case (a) can the PMTU data be maintained at the same
granularity as communication associations. In (b) and (c), the IP
layer will only be able to maintain PMTU data at the granularity of
source and destination IP addresses (and optionally TOS), as
described in RFC 1191. This is an important difference, because more
than one communication association may map to the same source and
destination IP addresses, and each communication association may have
a different amount of IPsec header overhead (e.g., due to use of
different transforms or different algorithms).

Implementation of the calculation of PMTU and support for PMTUs at
the granularity of individual communication associations is a local
matter. However, a socket-based implementation of IPsec in a host
SHOULD maintain the information on a per socket basis. Bump in the
stack systems MUST pass an ICMP PMTU to the host IP implementation,
after adjusting it for any IPsec header overhead added by these
systems. The calculation of the overhead SHOULD be determined by
analysis of the SPI and any other selector information present in a
returned ICMP PMTU message.

6.1.2.4 PMTU Aging

In all systems (host or gateway) implementing IPsec and maintaining
PMTU information, the PMTU associated with a security association
(transport or tunnel) MUST be "aged" and some mechanism put in place
for updating the PMTU in a timely manner, especially for discovering
if the PMTU is smaller than it needs to be. A given PMTU has to
remain in place long enough for a packet to get from the source end
of the security association to the system at the other end of the
security association and propagate back an ICMP error message if the
current PMTU is too big. Note that if there are nested tunnels,
multiple packets and round trip times might be required to get an
ICMP message back to an encapsulator or originating host.

Systems SHOULD use the approach described in the Path MTU Discovery
document (RFC 1191, Section 6.3), which suggests periodically
resetting the PMTU to the first-hop data-link MTU and then letting
the normal PMTU Discovery processes update the PMTU as necessary.
The period SHOULD be configurable.

7. Auditing

Not all systems that implement IPsec will implement auditing. For
the most part, the granularity of auditing is a local matter.
However, several auditable events are identified in the AH and ESP
specifications and for each of these events a minimum set of
information that SHOULD be included in an audit log is defined.
Additional information also MAY be included in the audit log for each
of these events, and additional events, not explicitly called out in
this specification, also MAY result in audit log entries. There is
no requirement for the receiver to transmit any message to the
purported transmitter in response to the detection of an auditable
event, because of the potential to induce denial of service via such
action.

8. Use in Systems Supporting Information Flow Security

Information of various sensitivity levels may be carried over a
single network. Information labels (e.g., Unclassified, Company
Proprietary, Secret) [DoD85, DoD87] are often employed to distinguish
such information. The use of labels facilitates segregation of
information, in support of information flow security models, e.g.,
the Bell-LaPadula model [BL73]. Such models, and corresponding
supporting technology, are designed to prevent the unauthorized flow
of sensitive information, even in the face of Trojan Horse attacks.
Conventional, discretionary access control (DAC) mechanisms, e.g.,
based on access control lists, generally are not sufficient to
support such policies, and thus facilities such as the SPD do not
suffice in such environments.

In the military context, technology that supports such models is
often referred to as multi-level security (MLS). Computers and
networks often are designated "multi-level secure" if they support
the separation of labelled data in conjunction with information flow
security policies. Although such technology is more broadly
applicable than just military applications, this document uses the
acronym "MLS" to designate the technology, consistent with much
extant literature.

IPsec mechanisms can easily support MLS networking. MLS networking
requires the use of strong Mandatory Access Controls (MAC), which
unprivileged users or unprivileged processes are incapable of
controlling or violating. This section pertains only to the use of
these IP security mechanisms in MLS (information flow security
policy) environments. Nothing in this section applies to systems not
claiming to provide MLS.

As used in this section, "sensitivity information" might include
implementation-defined hierarchic levels, categories, and/or
releasability information.

AH can be used to provide strong authentication in support of
mandatory access control decisions in MLS environments. If explicit
IP sensitivity information (e.g., IPSO [Ken91]) is used and
confidentiality is not considered necessary within the particular
operational environment, AH can be used to authenticate the binding
between sensitivity labels in the IP header and the IP payload
(including user data). This is a significant improvement over
labeled IPv4 networks where the sensitivity information is trusted
even though there is no authentication or cryptographic binding of
the information to the IP header and user data. IPv4 networks might
or might not use explicit labelling. IPv6 will normally use implicit
sensitivity information that is part of the IPsec Security
Association but not transmitted with each packet instead of using
explicit sensitivity information. All explicit IP sensitivity
information MUST be authenticated using either ESP, AH, or both.

Encryption is useful and can be desirable even when all of the hosts
are within a protected environment, for example, behind a firewall or
disjoint from any external connectivity. ESP can be used, in
conjunction with appropriate key management and encryption
algorithms, in support of both DAC and MAC. (The choice of
encryption and authentication algorithms, and the assurance level of
an IPsec implementation will determine the environments in which an
implementation may be deemed sufficient to satisfy MLS requirements.)
Key management can make use of sensitivity information to provide
MAC. IPsec implementations on systems claiming to provide MLS SHOULD
be capable of using IPsec to provide MAC for IP-based communications.

8.1 Relationship Between Security Associations and Data Sensitivity

Both the Encapsulating Security Pay load and the Authentication Header
can be combined with appropriate Security Association policies to
provide multi-level secure networking. In this case each SA (or SA
bundle) is normally used for only a single instance of sensitivity
information. For example, "PROPRIETARY - Internet Engineering" must
be associated with a different SA (or SA bundle) from "PROPRIETARY -
Finance".

8.2 Sensitivity Consistency Checking

An MLS implementation (both host and router) MAY associate
sensitivity information, or a range of sensitivity information with
an interface, or a configured IP address with its associated prefix
(the latter is sometimes referred to as a logical interface, or an

interface alias). If such properties exist, an implementation SHOULD
compare the sensitivity information associated with the packet
against the sensitivity information associated with the interface or
address/prefix from which the packet arrived, or through which the
packet will depart. This check will either verify that the
sensitivities match, or that the packet's sensitivity falls within
the range of the interface or address/prefix.

The checking SHOULD be done on both inbound and outbound processing.

8.3 Additional MLS Attributes for Security Association Databases

Section 4.4 discussed two Security Association databases (the
Security Policy Database (SPD) and the Security Association Database
(SAD)) and the associated policy selectors and SA attributes. MLS
networking introduces an additional selector/attribute:

- Sensitivity information.

The Sensitivity information aids in selecting the appropriate
algorithms and key strength, so that the traffic gets a level of
protection appropriate to its importance or sensitivity as described
in section 8.1. The exact syntax of the sensitivity information is
implementation defined.

8.4 Additional Inbound Processing Steps for MLS Networking

After an inbound packet has passed through IPsec processing, an MLS
implementation SHOULD first check the packet's sensitivity (as
defined by the SA (or SA bundle) used for the packet) with the
interface or address/prefix as described in section 8.2 before
delivering the datagram to an upper-layer protocol or forwarding it.

The MLS system MUST retain the binding between the data received in
an IPsec protected packet and the sensitivity information in the SA
or SAs used for processing, so appropriate policy decisions can be
made when delivering the datagram to an application or forwarding
engine. The means for maintaining this binding are implementation
specific.

8.5 Additional Outbound Processing Steps for MLS Networking

An MLS implementation of IPsec MUST perform two additional checks
besides the normal steps detailed in section 5.1.1. When consulting
the SPD or the SAD to find an outbound security association, the MLS
implementation MUST use the sensitivity of the data to select an

appropriate outbound SA or SA bundle. The second check comes before
forwarding the packet out to its destination, and is the sensitivity
consistency checking described in section 8.2.

8.6 Additional MLS Processing for Security Gateways

An MLS security gateway MUST follow the previously mentioned inbound
and outbound processing rules as well as perform some additional
processing specific to the intermediate protection of packets in an
MLS environment.

A security gateway MAY act as an outbound proxy, creating SAs for MLS
systems that originate packets forwarded by the gateway. These MLS
systems may explicitly label the packets to be forwarded, or the
whole originating network may have sensitivity characteristics
associated with it. The security gateway MUST create and use
appropriate SAs for AH, ESP, or both, to protect such traffic it
forwards.

Similarly such a gateway SHOULD accept and process inbound AH and/or
ESP packets and forward appropriately, using explicit packet
labeling, or relying on the sensitivity characteristics of the
destination network.

9. Performance Issues

The use of IPsec imposes computational performance costs on the hosts
or security gateways that implement these protocols. These costs are
associated with the memory needed for IPsec code and data structures,
and the computation of integrity check values, encryption and
decryption, and added per-packet handling. The per-packet
computational costs will be manifested by increased latency and,
possibly, reduced throughout. Use of SA/key management protocols,
especially ones that employ public key cryptography, also adds
computational performance costs to use of IPsec. These per-
association computational costs will be manifested in terms of
increased latency in association establishment. For many hosts, it
is anticipated that software-based cryptography will not appreciably
reduce throughput, but hardware may be required for security gateways
(since they represent aggregation points), and for some hosts.

The use of IPsec also imposes bandwidth utilization costs on
transmission, switching, and routing components of the Internet
infrastructure, components not implementing IPsec. This is due to
the increase in the packet size resulting from the addition of AH
and/or ESP headers, AH and ESP tunneling (which adds a second IP
header), and the increased packet traffic associated with key
management protocols. It is anticipated that, in most instances,

this increased bandwidth demand will not noticeably affect the
Internet infrastructure. However, in some instances, the effects may
be significant, e.g., transmission of ESP encrypted traffic over a
dialup link that otherwise would have compressed the traffic.

Note: The initial SA establishment overhead will be felt in the first
packet. This delay could impact the transport layer and application.
For example, it could cause TCP to retransmit the SYN before the
ISAKMP exchange is done. The effect of the delay would be different
on UDP than TCP because TCP shouldn't transmit anything other than
the SYN until the connection is set up whereas UDP will go ahead and
transmit data beyond the first packet.

Note: As discussed earlier, compression can still be employed at
layers above IP. There is an IETF working group (IP Payload
Compression Protocol (ippcp)) working on "protocol specifications
that make it possible to perform lossless compression on individual
payloads before the payload is processed by a protocol that encrypts
it. These specifications will allow for compression operations to be
performed prior to the encryption of a payload by IPsec protocols."

10. Conformance Requirements

All IPv4 systems that claim to implement IPsec MUST comply with all
requirements of the Security Architecture document. All IPv6 systems
MUST comply with all requirements of the Security Architecture
document.

11. Security Considerations

The focus of this document is security; hence security considerations
permeate this specification.

12. Differences from RFC 1825

This architecture document differs substantially from RFC 1825 in
detail and in organization, but the fundamental notions are
unchanged. This document provid es considerable additional detail in
terms of compliance specifications. It introduces the SPD and SAD,
and the notion of SA selectors. It is aligned with the new versions
of AH and ESP, which also differ from their predecessors. Specific
requirements for supported combinations of AH and ESP are newly
added, as are details of PMTU management.

Acknowledgements

Many of the concepts embodied in this specification were derived from
or influenced by the US Government's SP3 security protocol, ISO/IEC's
NLSP, the proposed swIPe security protocol [SDNS, ISO, IB93, IBK93],
and the work done for SNMP Security and SNMPv2 Security.

For over 3 years (although it sometimes seems *much* longer), this
document has evolved through multiple versions and iterations.
During this time, many people have contributed significant ideas and
energy to the process and the documents themselves. The authors
would like to thank Karen Seo for providing extensive help in the
review, editing, background research, and coordination for this
version of the specification. The authors would also like to thank
the members of the IPsec and IPng working groups, with special
mention of the efforts of (in alphabetic order): Steve Bellovin,
Steve Deering, James Hughes, Phil Karn, Frank Kastenholz, Perry
Metzger, David Mihelcic, Hilarie Orman, Norman Shulman, William
Simpson, Harry Varnis, and Nina Yuan.

Appendix A -- Glossary

This section provides definitions for several key terms that are
employed in this document. Other documents provide additional
definitions and background information relevant to this technology,
e.g., [VK83, HA94]. Included in this glossary are generic security
service and security mechanism terms, plus IPsec-specific terms.

Access Control
Access control is a security service that prevents unauthorized
use of a resource, including the prevention of use of a resource
in an unauthorized manner. In the IPsec context, the resource
to which access is being controlled is often:
o for a host, computing cycles or data
o for a security gateway, a network behind the gateway
or
bandwidth on that network.

Anti-replay
[See "Integrity" below]

Authentication
This term is used informally to refer to the combination of two
nominally distinct security services, data origin authentication
and connectionless integrity. See the definitions below for
each of these services.

Availability
Availability, when viewed as a security service, addresses the
security concerns engendered by attacks against networks that
deny or degrade service. For example, in the IPsec context, the
use of anti-replay mechanisms in AH and ESP support
availability.

Confidentiality
Confidentiality is the security service that protects data from
unauthorized disclosure. The primary confidentiality concern in
most instances is unauthorized disclosure of application level
data, but disclosure of the external characteristics of
communication also can be a concern in some circumstances.
Traffic flow confidentiality is the service that addresses this
latter concern by concealing source and destination addresses,
message length, or frequency of communication. In the IPsec
context, using ESP in tunnel mode, especially at a security
gateway, can provide some level of traffic flow confidentiality.
(See also traffic analysis, below.)

Encryption
Encryption is a security mechanism used to transform data from
an intelligible form (plaintext) into an unintelligible form
(ciphertext), to provide confidentiality. The inverse
transformation process is designated "decryption". Oftimes the
term "encryption" is used to generically refer to both
processes.

Data Origin Authentication
Data origin authentication is a security service that verifies
the identity of the claimed source of data. This service is
usually bundled with connectionless integrity service.

Integrity
Integrity is a security service that ensures that modifications
to data are detectable. Integrity comes in various flavors to
match application requirements. IPsec supports two forms of
integrity: connectionless and a form of partial sequence
integrity. Connectionless integrity is a service that detects
modification of an individual IP datagram, without regard to the
ordering of the datagram in a stream of traffic. The form of
partial sequence integrity offered in IPsec is referred to as
anti-replay integrity, and it detects arrival of duplicate IP
datagrams (within a constrained window). This is in contrast to
connection-oriented integrity, which imposes more stringent
sequencing requirements on traffic, e.g., to be able to detect
lost or re-ordered messages. Although authentication and
integrity services often are cited separately, in practice they
are intimately connected and almost always offered in tandem.

Security Association (SA)
A simplex (uni-directional) logical connection, created for
security purposes. All traffic traversing an SA is provided the
same security processing. In IPsec, an SA is an internet layer
abstraction implemented through the use of AH or ESP.

Security Gateway
A security gateway is an intermediate system that acts as the
communications interface between two networks. The set of hosts
(and networks) on the external side of the security gateway is
viewed as untrusted (or less trusted), while the networks and
hosts and on the internal side are viewed as trusted (or more
trusted). The internal subnets and hosts served by a security
gateway are presumed to be trusted by virtue of sharing a
common, local, security administration. (See "Trusted
Subnetwork" below.) In the IPsec context, a security gateway is
a point at which AH and/or ESP is implemented in order to serve

a set of internal hosts, providing security services for these
hosts when they communicate with external hosts also employing
IPsec (either directly or via another security gateway).

SPI
Acronym for "Security Parameters Index". The combination of a
destination address, a security protocol, and an SPI uniquely
identifies a security association (SA, see above). The SPI is
carried in AH and ESP protocols to enable the receiving system
to select the SA under which a received packet will be
processed. An SPI has only local significance, as defined by
the creator of the SA (usually the receiver of the packet
carrying the SPI); thus an SPI is generally viewed as an opaque
bit string. However, the creator of an SA may choose to
interpret the bits in an SPI to facilitate local processing.

Traffic Analysis
The analysis of network traffic flow for the purpose of deducing
information that is useful to an adversary. Examples of such
informat ion are frequency of transmission, the identities of the
conversing parties, sizes of packets, flow identifiers, etc.
[Sch94]

Trusted Subnetwork
A subnetwork containing hosts and routers that trust each other
not to engage in active or passive attacks. There also is an
assumption that the underlying communications channel (e.g., a
LAN or CAN) isn't being attacked by other means.

Appendix B -- Analysis/Discussion of PMTU/DF/Fragmentation Issues

B.1 DF bit

In cases where a system (host or gateway) adds an encapsulating
header (e.g., ESP tunnel), should/must the DF bit in the original
packet be copied to the encapsulating header?

Fragmenting seems correct for some situations, e.g., it might be
appropriate to fragment packets over a network with a very small MTU,
e.g., a packet radio network, or a cellular phone hop to mobile node,
rather than propagate back a very small PMTU for use over the rest of
the path. In other situations, it might be appropriate to set the DF
bit in order to get feedback from later routers about PMTU
constraints which require fragmentation. The existence of both of
these situations argues for enabling a system to decide whether or
not to fragment over a particular network "link", i.e., for requiring
an implementation to be able to copy the DF bit (and to process ICMP
PMTU messages), but making it an option to be selected on a per
interface basis. In other words, an administrator should be able to
configure the router's treatment of the DF bit (set, clear, copy from
encapsulated header) for each interface.

Note: If a bump-in-the-stack implementation of IPsec attempts to
apply different IPsec algorithms based on source/destination ports,
it will be difficult to apply Path MTU adjustments.

B.2 Fragmentation

If required, IP fragmentation occurs after IPsec processing within an
IPsec implementation. Thus, transport mode AH or ESP is applied only
to whole IP datagrams (not to IP fragments). An IP packet to which
AH or ESP has been applied may itself be fragmented by routers en
route, and such fragments MUST be reassembled prior to IPsec
processing at a receiver. In tunnel mode, AH or ESP is applied to an
IP packet, the payload of which may be a fragmented IP packet. For
example, a security gateway, "bump-in-the-stack" (BITS), or "bump-
in-the-wire" (BITW) IPsec implementation may apply tunnel mode AH to
such fragments. Note that BITS or BITW implementations are examples
of where a host IPsec implementation might receive fragments to which
tunnel mode is to be applied. However, if transport mode is to be
applied, then these implementations MUST reassemble the fragments
prior to applying IPsec.

NOTE: IPsec always has to figure out what the encapsulating IP header
fields are. This is independent of where you insert IPsec and is
intrinsic to the definition of IPsec. Therefore any IPsec
implementation that is not integrated into an IP implementation must
include code to construct the necessary IP headers (e.g., IP2):

o AH-tunnel --> IP2-AH-IP1-Transport-Data
o ESP-tunnel --> IP2-ESP_hdr-IP1-Transport-Data-ESP_trailer

*********************************************************************

Overall, the fragmentation/reassembly approach described above works
for all cases examined.

AH Xport AH Tunnel ESP Xport ESP Tunnel
Implementation approach IPv4 IPv6 IPv4 IPv6 IPv4 IPv6 IPv4 IPv6
----------------------- ---- ---- ---- ---- ---- ---- ---- ----
Hosts (integr w/ IP stack) Y Y Y Y Y Y Y Y
Hosts (betw/ IP and drivers) Y Y Y Y Y Y Y Y
S. Gwy (integr w/ IP stack) Y Y Y Y
Outboard crypto processor *

* If the crypto processor system has its own IP address, then it
is covered by the security gateway case. This box receives
the packet from the host and performs IPsec processing. It
has to be able to handle the same AH, ESP, and related
IPv4/IPv6 tunnel processing that a security gateway would have
to handle. If it doesn't have it's own address, then it is
similar to the bump-in-the stack implementation between IP and
the network drivers.

The following analysis assumes that:

1. There is only one IPsec module in a given system's stack.
There isn't an IPsec module A (adding ESP/encryption and
thus) hiding the transport protocol, SRC port, and DEST port
from IPsec module B.
2. There are several places where IPsec could be implemented (as
shown in the table above).
a. Hosts with integration of IPsec into the native IP
implementation. Implementer has access to the source
for the stack.
b. Hosts with bump-in-the-stack implementations, where
IPsec is implemented between IP and the local network
drivers. Source access for stack is not available;
but there are well-defined interfaces that allows the
IPsec code to be incorporated into the system.

c. Security gateways and outboard crypto processors with
integration of IPsec into the stack.
3. Not all of the above approaches are feasible in all hosts.
But it was assumed that for each approach, there are some
hosts for whom the approach is feasible.

For each of the above 3 categories, there are IPv4 and IPv6, AH
transport and tunnel modes, and ESP transport and tunnel modes -- for
a total of 24 cases (3 x 2 x 4).

Some header fields and interface fields are listed here for ease of
reference -- they're not in the header order, but instead listed to
allow comparison between the columns. (* = not covered by AH
authentication. ESP authentication doesn't cover any headers that
precede it.)

IP/Transport Interface
IPv4 IPv6 (RFC 1122 -- Sec 3.4)
---- ---- ----------------------
Version = 4 Version = 6
Header Len
*TOS Class,Flow Lbl TOS
Packet Len Payload Len Len
ID ID (optional)
*Flags DF
*Offset
*TTL *Hop Limit TTL
Protocol Next Header
*Checksum
Src Address Src Address Src Address
Dst Address Dst Address Dst Address
Options? Options? Opt

? = AH covers Option-Type and Option-Length, but
might not cover Option-Data.

The results for each of the 20 cases is shown below ("works" = will
work if system fragments after outbound IPsec processing, reassembles
before inbound IPsec processing). Notes indicate implementation
issues.

a. Hosts (integrated into IP stack)
o AH-transport --> (IP1-AH-Transport-Data)
- IPv4 -- works
- IPv6 -- works
o AH-tunnel --> (IP2-AH-IP1-Transport-Data)
- IPv4 -- works
- IPv6 -- works

o ESP-transport --> (IP1-ESP_hdr-Transport-Data-ESP_trailer)
- IPv4 -- works
- IPv6 -- works
o ESP-tunnel --> (IP2-ESP_hdr-IP1-Transport-Data-ESP_trailer)
- IPv4 -- works
- IPv6 -- works

b. Hosts (Bump-in-the-stack) -- put IPsec between IP layer and
network drivers. In this case, the IPsec module would have to do
something like one of the following for fragmentation and
reassembly.
- do the fragmentation/reassembly work itself and
send/receive the packet directly to/from the network
layer. In AH or ESP transport mode, this is fine. In AH
or ESP tunnel mode where the tunnel end is at the ultimate
destination, this is fine. But in AH or ESP tunnel modes
where the tunnel end is different from the ultimate
destination and where the source host is multi-homed, this
approach could result in sub-optimal routing because the
IPsec module may be unable to obtain the information
needed (LAN interface and next-hop gateway) to direct the
packet to the appropriate network interface. This is not
a problem if the interface and next-hop gateway are the
same for the ultimate destination and for the tunnel end.
But if they are different, then IPsec would need to know
the LAN interface and the next-hop gateway for the tunnel
end. (Note: The tunnel end (security gateway) is highly
likely to be on the regular path to the ultimate
destination. But there could also be more than one path
to the destination, e.g., the host could be at an
organization with 2 firewalls. And the path being used
could involve the less commonly chosen firewall.) OR
- pass the IPsec'd packet back to the IP layer where an
extra IP header would end up being pre-pended and the
IPsec module would have to check and let IPsec'd fragments
go by.
OR
- pass the packet contents to the IP layer in a form such
that the IP layer recreates an appropriate IP header

At the network layer, the IPsec module will have access to the
following selectors from the packet -- SRC address, DST address,
Next Protocol, and if there's a transport layer header --> SRC
port and DST port. One cannot assume IPsec has access to the
Name. It is assumed that the available selector information is
sufficient to figure out the relevant Security Policy entry and
Security Association(s).

o AH-transport --> (IP1-AH-Transport-Data)
- IPv4 -- works
- IPv6 -- works
o AH-tunnel --> (IP2-AH-IP1-Transport-Data)
- IPv4 -- works
- IPv6 -- works
o ESP-transport --> (IP1-ESP_hdr-Transport-Data-ESP_trailer)
- IPv4 -- works
- IPv6 -- works
o ESP-tunnel --> (IP2-ESP_hdr-IP1-Transport-Data-ESP_trailer)
- IPv4 -- works
- IPv6 -- works

c. Security gateways -- integrate IPsec into the IP stack

NOTE: The IPsec module will have access to the following
selectors from the packet -- SRC address, DST address, Next
Protocol, and if there's a transport layer header --> SRC port
and DST port. It won't have access to the User ID (only Hosts
have access to User ID information.) Unlike some Bump-in-the-
stack implementations, security gateways may be able to look up
the Source Address in the DNS to provide a System Name, e.g., in
situations involving use of dynamically assigned IP addresses in
conjunction with dynamically updated DNS entries. It also won't
have access to the transport layer information if there is an ESP
header, or if it's not the first fragment of a fragmented
message. It is assumed that the available selector information
is sufficient to figure out the relevant Security Policy entry
and Security Association(s).

o AH-tunnel --> (IP2-AH-IP1-Transport-Data)
- IPv4 -- works
- IPv6 -- works
o ESP-tunnel --> (IP2-ESP_hdr-IP1-Transport-Data-ESP_trailer)
- IPv4 -- works
- IPv6 -- works

**********************************************************************

B.3 Path MTU Discovery

As mentioned earlier, "ICMP PMTU" refers to an ICMP message used for
Path MTU Discovery.

The legend for the diagrams below in B.3.1 and B.3.3 (but not B.3.2)
is:

==== = security association (AH or ESP, transport or tunnel)

---- = connectivity (or if so labelled, administrative boundary)
.... = ICMP message (hereafter referred to as ICMP PMTU) for

IPv4:
- Type = 3 (Destination Unreachable)
- Code = 4 (Fragmentation needed and DF set)
- Next-Hop MTU in the low-order 16 bits of the second
word of the ICMP header (labelled unused in RFC 792),
with high-order 16 bits set to zero

IPv6 (RFC 1885):
- Type = 2 (Packet Too Big)
- Code = 0 (Fragmentation needed and DF set)
- Next-Hop MTU in the 32 bit MTU field of the ICMP6

Hx = host x
Rx = router x
SGx = security gateway x
X* = X supports IPsec

B.3.1 Identifying the Originating Host(s)

The amount of information returned with the ICMP message is limited
and this affects what selectors are available to identify security
associations, originating hosts, etc. for use in further propagating
the PMTU information.

In brief... An ICMP message must contain the following information
from the "offending" packet:
- IPv4 (RFC 792) -- IP header plus a minimum of 64 bits

Accordingly, in the IPv4 context, an ICMP PMTU may identify only the
first (outermost) security association. This is because the ICMP
PMTU may contain only 64 bits of the "offending" packet beyond the IP
header, which would capture only the first SPI from AH or ESP. In
the IPv6 context, an ICMP PMTU will probably provide all the SPIs and
the selectors in the IP header, but maybe not the SRC/DST ports (in
the transport header) or the encapsulated (TCP, UDP, etc.) protocol.
Moreover, if ESP is used, the transport ports and protocol selectors
may be encrypted.

Looking at the diagram below of a security gateway tunnel (as
mentioned elsewhere, security gateways do not use transport mode)...

H1 =================== H3
\ | | /
H0 -- SG1* ---- R1 ---- SG2* ---- R2 -- H5
/ ^ | \
H2 |........| H4

Suppose that the security policy for SG1 is to use a single SA to SG2
for all the traffic between hosts H0, H1, and H2 and hosts H3, H4,
and H5. And suppose H0 sends a data packet to H5 which causes R1 to
send an ICMP PMTU message to SG1. If the PMTU message has only the
SPI, SG1 will be able to look up the SA and find the list of possible
hosts (H0, H1, H2, wildcard); but SG1 will have no way to figure out
that H0 sent the traffic that triggered the ICMP PMTU message.

original after IPsec ICMP
packet processing packet
-------- ----------- ------
IP-3 header (S = R1, D = SG1)
ICMP header (includes PMTU)
IP-2 header IP-2 header (S = SG1, D = SG2)
ESP header minimum of 64 bits of ESP hdr (*)
IP-1 header IP-1 header
TCP header TCP header
TCP data TCP data
ESP trailer

(*) The 64 bits will include enough of the ESP (or AH) header to
include the SPI.
- ESP -- SPI (32 bits), Seq number (32 bits)
- AH -- Next header (8 bits), Payload Len (8 bits),
Reserved (16 bits), SPI (32 bits)

This limitation on the amount of information returned with an ICMP
message creates a problem in identifying the originating hosts for
the packet (so as to know where to further propagate the ICMP PMTU
information). If the ICMP message contains only 64 bits of the IPsec
header (minimum for IPv4), then the IPsec selectors (e.g., Source and
Destination addresses, Next Protocol, Source and Destination ports,
etc.) will have been lost. But the ICMP error message will still
provide SG1 with the SPI, the PMTU information and the source and
destination gateways for the relevant security association.

The destination security gateway and SPI uniquely define a security
association which in turn defines a set of possible originating
hosts. At this point, SG1 could:

a. send the PMTU information to all the possible originating hosts.
This would not work well if the host list is a wild card or if
many/most of the hosts weren't sending to SG1; but it might work
if the SPI/destination/etc mapped to just one or a small number of
hosts.
b. store the PMTU with the SPI/etc and wait until the next packet(s)
arrive from the originating host(s) for the relevant security
association. If it/they are bigger than the PMTU, drop the
packet(s), and compose ICMP PMTU message(s) with the new packet(s)
and the updated PMTU, and send the originating host(s) the ICMP
message(s) about the problem. This involves a delay in notifying
the originating host(s), but avoids the problems of (a).

Since only the latter approach is feasible in all instances, a
security gateway MUST provide such support, as an option. However,
if the ICMP message contains more information from the original
packet, then there may be enough information to immediately determine
to which host to propagate the ICMP/PMTU message and to provide that
system with the 5 fields (source address, destination address, source
port, destination port, and transport protocol) needed to determine
where to store/update the PMTU. Under such circumstances, a security
gateway MUST generate an ICMP PMTU message immediately upon receipt
of an ICMP PMTU from further down the path. NOTE: The Next Protocol
field may not be contained in the ICMP message and the use of ESP
encryption may hide the selector fields that have been encrypted.

B.3.2 Calculation of PMTU

The calculation of PMTU from an ICMP PMTU has to take into account
the addition of any IPsec header by H1 -- AH and/or ESP transport, or
ESP or AH tunnel. Within a single host, multiple applications may
share an SPI and nesting of security associations may occur. (See
Section 4.5 Basic Combinations of Security Associations for
description of the combinations that MUST be supported). The diagram
below illustrates an example of security associations between a pair
of hosts (as viewed from the perspective of one of the hosts.) (ESPx
or AHx = transport mode)

Socket 1 -------------------------|
|
Socket 2 (ESPx/SPI-A) ---------- AHx (SPI-B) -- Internet

In order to figure out the PMTU for each socket that maps to SPI-B,
it will be necessary to have backpointers from SPI-B to each of the 2
paths that lead to it -- Socket 1 and Socket 2/SPI-A.

B.3.3 Granularity of Maintaining PMTU Data

In hosts, the granularity with which PMTU ICMP processing can be done
differs depending on the implementation situation. Looking at a
host, there are three situations that are of interest with respect to
PMTU issues:

a. Integration of IPsec into the native IP implementation
b. Bump-in-the-stack implementations, where IPsec is implemented
"underneath" an existing implementation of a TCP/IP protocol
stack, between the native IP and the local network drivers
c. No IPsec implementation -- This case is included because it is
relevant in cases where a security gateway is sending PMTU
information back to a host.

Only in case (a) can the PMTU data be maintained at the same
granularity as communication associations. In the other cases, the
IP layer will maintain PMTU data at the granularity of Source and
Destination IP addresses (and optionally TOS/Class), as described in
RFC 1191. This is an important difference, because more than one
communication association may map to the same source and destination
IP addresses, and each communication association may have a different
amount of IPsec header overhead (e.g., due to use of different
transforms or different algorithms). The examples below illustrate
this.

In cases (a) and (b)... Suppose you have the following situation.
H1 is sending to H2 and the packet to be sent from R1 to R2 exceeds
the PMTU of the network hop between them.

==================================
| |
H1* --- R1 ----- R2 ---- R3 ---- H2*
^ |
|.......|

If R1 is configured to not fragment subscriber traffic, then R1 sends
an ICMP PMTU message with the appropriate PMTU to H1. H1's
processing would vary with the nature of the implementation. In case
(a) (native IP), the security services are bound to sockets or the
equivalent. Here the IP/IPsec implementation in H1 can store/update
the PMTU for the associated socket. In case (b), the IP layer in H1
can store/update the PMTU but only at the granularity of Source and
Destination addresses and possibly TOS/Class, as noted above. So the
result may be sub-optimal, since the PMTU for a given
SRC/DST/TOS/Class will be the subtraction of the largest amount of
IPsec header used for any communication association between a given
source and destination.

In case (c), there has to be a security gateway to have any IPsec
processing. So suppose you have the following situation. H1 is
sending to H2 and the packet to be sent from SG1 to R exceeds the
PMTU of the network hop between them.

================
| |
H1 ---- SG1* --- R --- SG2* ---- H2
^ |
|.......|

As described above for case (b), the IP layer in H1 can store/update
the PMTU but only at the granularity of Source and Destination
addresses, and possibly TOS/Class. So the result may be sub-optimal,
since the PMTU for a given SRC/DST/TOS/Class will be the subtraction
of the largest amount of IPsec header used for any communication
association between a given source and destination.

B.3.4 Per Socket Maintenance of PMTU Data

Implementation of the calculation of PMTU (Section B.3.2) and support
for PMTUs at the granularity of individual "communication
associations" (Section B.3.3) is a local matter. However, a socket-
based implementation of IPsec in a host SHOULD maintain the
information on a per socket basis. Bump in the stack systems MUST
pass an ICMP PMTU to the host IP implementation, after adjusting it
for any IPsec header overhead added by these systems. The
determination of the overhead SHOULD be determined by analysis of the
SPI and any other selector information present in a returned ICMP
PMTU message.

B.3.5 Delivery of PMTU Data to the Transport Layer

The host mechanism for getting the updated PMTU to the transport
layer is unchanged, as specified in RFC 1191 (Path MTU Discovery).

B.3.6 Aging of PMTU Data

This topic is covered in Section 6.1.2.4.

Appendix C -- Sequence Space Window Code Example

This appendix contains a routine that implements a bitmask check for
a 32 packet window. It was provided by James Hughes
(jim_hughes@stortek.com) and Harry Varnis (hgv@anubis.network.com)
and is intended as an implementation example. Note that this code
both checks for a replay and updates the window. Thus the algorithm,
as shown, should only be called AFTER the packet has been
authenticated. Implementers might wish to consider splitting the
code to do the check for replays before computing the ICV. If the
packet is not a replay, the code would then compute the ICV, (discard
any bad packets), and if the packet is OK, update the window.

#include <stdio.h>
#include <stdlib.h>
typedef unsigned long u_long;

enum {
ReplayWindowSize = 32
};

u_long bitmap = 0; /* session state - must be 32 bits */
u_long lastSeq = 0; /* session state */

/* Returns 0 if packet disallowed, 1 if packet permitted */
int ChkReplayWindow(u_long seq);

int ChkReplayWindow(u_long seq) {
u_long diff;

if (seq == 0) return 0; /* first == 0 or wrapped */
if (seq > lastSeq) { /* new larger sequence number */
diff = seq - lastSeq;
if (diff < ReplayWindowSize) { /* In window */
bitmap <<= diff;
bitmap |= 1; /* set bit for this packet */
} else bitmap = 1; /* This packet has a "way larger" */
lastSeq = seq;
return 1; /* larger is good */
}
diff = lastSeq - seq;
if (diff >= ReplayWindowSize) return 0; /* too old or wrapped */
if (bitmap & ((u_long)1 << diff)) return 0; /* already seen */
bitmap |= ((u_long)1 << diff); /* mark as seen */
return 1; /* out of order but good */
}

char string_buffer[512];

#define STRING_BUFFER_SIZE sizeof(string_buffer)

int main() {
int result;
u_long last, current, bits;

printf("Input initial state (bits in hex, last msgnum):\n");
if (!fgets(string_buffer, STRING_BUFFER_SIZE, stdin)) exit(0);
sscanf(string_buffer, "%lx %lu", &bits, &last);
if (last != 0)
bits |= 1;
bitmap = bits;
lastSeq = last;
printf("bits:%08lx last:%lu\n", bitmap, lastSeq);
printf("Input value to test (current):\n");

while (1) {
if (!fgets(string_buffer, STRING_BUFFER_SIZE, stdin)) break;
sscanf(string_buffer, "%lu", ¤t);
result = ChkReplayWindow(current);
printf("%-3s", result ? "OK" : "BAD");
printf(" bits:%08lx last:%lu\n", bitmap, lastSeq);
}
return 0;
}

Appendix D -- Categorization of ICMP messages

The tables below characterize ICMP messages as being either host
generated, router generated, both, unassigned/unknown. The first set
are IPv4. The second set are IPv6.

IPv4

Type Name/Codes Reference
========================================================================
HOST GENERATED:
3 Destination Unreachable
2 Protocol Unreachable [RFC792]
3 Port Unreachable [RFC792]
8 Source Host Isolated [RFC792]
14 Host Precedence Violation [RFC1812]
10 Router Selection [RFC1256]

Type Name/Codes Reference
========================================================================
ROUTER GENERATED:
3 Destination Unreachable
0 Net Unreachable [RFC792]
4 Fragmentation Needed, Don't Fragment was Set [RFC792]
5 Source Route Failed [RFC792]
6 Destination Network Unknown [RFC792]
7 Destination Host Unknown [RFC792]
9 Comm. w/Dest. Net. is Administratively Prohibited [RFC792]
11 Destination Network Unreachable for Type of Service[RFC792]
5 Redirect
0 Redirect Datagram for the Network (or subnet) [RFC792]
2 Redirect Datagram for the Type of Service & Network[RFC792]
9 Router Advertisement [RFC1256]
18 Address Mask Reply [RFC950]

IPv4
Type Name/Codes Reference
========================================================================
BOTH ROUTER AND HOST GENERATED:
0 Echo Reply [RFC792]
3 Destination Unreachable
1 Host Unreachable [RFC792]
10 Comm. w/Dest. Host is Administratively Prohibited [RFC792]
12 Destination Host Unreachable for Type of Service [RFC792]
13 Communication Administratively Prohibited [RFC1812]
15 Precedence cutoff in effect [RFC1812]
4 Source Quench [RFC792]
5 Redirect
1 Redirect Datagram for the Host [RFC792]
3 Redirect Datagram for the Type of Service and Host [RFC792]
6 Alternate Host Address [JBP]
8 Echo [RFC792]
11 Time Exceeded [RFC792]
12 Parameter Problem [RFC792,RFC1108]
13 Timestamp [RFC792]
14 Timestamp Reply [RFC792]
15 Information Re quest [RFC792]
16 Information Reply [RFC792]
17 Address Mask Request [RFC950]
30 Traceroute [RFC1393]
31 Datagram Conversion Error [RFC1475]
32 Mobile Host Redirect [Johnson]
39 SKIP [Markson]
40 Photuris [Simpson]

Type Name/Codes Reference
========================================================================
UNASSIGNED TYPE OR UNKNOWN GENERATOR:
1 Unassigned [JBP]
2 Unassigned [JBP]
7 Unassigned [JBP]
19 Reserved (for Security) [Solo]
20-29 Reserved (for Robustness Experiment) [ZSu]
33 IPv6 Where-Are-You [Simpson]
34 IPv6 I-Am-Here [Simpson]
35 Mobile Registration Request [Simpson]
36 Mobile Registration Reply [Simpson]
37 Domain Name Request [Simpson]
38 Domain Name Reply [Simpson]
41-255 Reserved [JBP]

IPv6

Type Name/Codes Reference
========================================================================
HOST GENERATED:
1 Destination Unreachable [RFC 1885]
4 Port Unreachable

Type Name/Codes Reference
========================================================================
ROUTER GENERATED:
1 Destination Unreachable [RFC1885]
0 No Route to Destination
1 Comm. w/Destination is Administratively Prohibited
2 Not a Neighbor
3 Address Unreachable
2 Packet Too Big [RFC1885]
0
3 Time Exceeded [RFC1885]
0 Hop Limit Exceeded in Transit
1 Fragment reassembly time exceeded

Type Name/Codes Reference
========================================================================
BOTH ROUTER AND HOST GENERATED:
4 Parameter Problem [RFC1885]
0 Erroneous Header Field Encountered
1 Unrecognized Next Header Type Encountered
2 Unrecognized IPv6 Option Encountered

References

[BL73] Bell, D.E. & LaPadula, L.J., "Secure Computer Systems:
Mathematical Foundations and Model", Technical Report M74-
244, The MITRE Corporation, Bedford, MA, May 1973.

[Bra97] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Level", BCP 14, RFC 2119, March 1997.

[DoD85] US National Computer Security Center, "Department of
Defense Trusted Computer System Evaluation Criteria", DoD
5200.28-STD, US Department of Defense, Ft. Meade, MD.,
December 1985.

[DoD87] US National Computer Security Center, "Trusted Network
Interpretation of the Trusted Computer System Evaluation
Criteria", NCSC-TG-005, Version 1, US Department of
Defense, Ft. Meade, MD., 31 July 1987.

[HA94] Haller, N., and R. Atkinson, "On Internet Authentication",
RFC 1704, October 1994.

[HC98] Harkins, D., and D. Carrel, "The Internet Key Exchange
(IKE)", RFC 2409, November 1998.

[HM97] Harney, H., and C. Muckenhirn, "Group Key Management
Protocol (GKMP) Architecture", RFC 2094, July 1997.

[ISO] ISO/IEC JTC1/SC6, Network Layer Security Protocol, ISO-IEC
DIS 11577, International Standards Organisation, Geneva,
Switzerland, 29 November 1992.

[IB93] John Ioannidis and Matt Blaze, "Architecture and
Implementation of Network-layer Security Under Unix",
Proceedings of USENIX Security Symposium, Santa Clara, CA,
October 1993.

[IBK93] John Ioannidis, Matt Blaze, & Phil Karn, "swIPe: Network-
Layer Security for IP", presentation at the Spring 1993
IETF Meeting, Columbus, Ohio

[KA98a] Kent, S., and R. Atkinson, "IP Authentication Header", RFC
2402, November 1998.

[KA98b] Kent, S., and R. Atkinson, "IP Encapsulating Security
Payload (ESP)", RFC 2406, November 1998.

[Ken91] Kent, S., "US DoD Security Options for the Internet
Protocol", RFC 1108, November 1991.

[MSST97] Maughan, D., Schertler, M., Schneider, M., and J. Turner,
"Internet Security Association and Key Management Protocol
(ISAKMP)", RFC 2408, November 1998.

[Orm97] Orman, H., "The OAKLEY Key Determination Protocol", RFC
2412, November 1998.

[Pip98] Piper, D., "The Internet IP Security Domain of
Interpretation for ISAKMP", RFC 2407, November 1998.

[Sch94] Bruce Schneier, Applied Cryptography, Section 8.6, John
Wiley & Sons, New York, NY, 1994.

[SDNS] SDNS Secure Data Network System, Security Protocol 3, SP3,
Document SDN.301, Revision 1.5, 15 May 1989, published in
NIST Publication NIST-IR-90-4250, February 1990.

[SMPT98] Shacham, A., Monsour, R., Pereira, R., and M. Thomas, "IP
Payload Compression Protocol (IPComp)", RFC 2393, August
1998.

[TDG97] Thayer, R., Doraswamy, N., and R. Glenn, "IP Security
Document Roadmap", RFC 2411, November 1998.

[VK83] V.L. Voydock & S.T. Kent, "Security Mechanisms in High-
level Networks", ACM Computing Surveys, Vol. 15, No. 2,
June 1983.

Disclaimer

The views and specification expressed in this document are those of
the authors and are not necessarily those of their employers. The
authors and their employers specifically disclaim responsibility for
any problems arising from correct or incorrect implementation or use
of this design.

Author Information

Stephen Kent
BBN Corporation
70 Fawcett Street
Cambridge, MA 02140
USA

Phone: +1 (617) 873-3988
EMail: kent@bbn.com

Randall Atkinson
@Home Network
425 Broadway
Redwood City, CA 94063
USA

Phone: +1 (415) 569-5000
EMail: rja@corp.home.net

Copyright (C) The Internet Society (1998). 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 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 a nd 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
English.

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
"AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.