Network Working Group P. Srisuresh
Request for Comments: 2709 LUCent Technologies
Category: Informational October 1999

Security Model with Tunnel-mode IPsec for NAT Domains

Status of this Memo

This memo provides information for the Internet community. It does
not specify an Internet standard of any kind. Distribution of this
memo is unlimited.

Copyright Notice

Copyright (C) The Internet Society (1999). All Rights Reserved.

Abstract

There are a variety of NAT flavors, as described in [Ref 1]. Of the
domains supported by NATs, only Realm-Specific IP clients are able to
pursue end-to-end IPsec secure sessions. However, all flavors of NAT
are capable of offering tunnel-mode IPsec security to private domain
hosts peering with nodes in external realm. This document describes a
security model by which tunnel-mode IPsec security can be architected
on NAT devices. A section is devoted to describing how security
policies may be transparently communicated to IKE (for automated KEY
exchange) during Quick Mode. Also outlined are applications that can
benefit from the Security Model described.

1. Introduction and Overview

NAT devices provide transparent routing to end hosts trying to
communicate from disparate address realms, by modifying IP and
transport headers en-route. This solution works best when the end
user identifier (such as host name) is different from the address
used to locate end user.

End-to-end application level payload security can be provided for
applications that do not embed realm-specific information in payloads
that is meaningless to one of the end-users. Applications that do
embed realm-specific information in payload will require an
application level gateway (ALG) to make the payload meaningful in
both realms. However, applications that require assistance of an ALG
en-route cannot pursue end-to-end application level security.

All applications traversing a NAT device, irrespective of whether
they require assistance of an ALG or not, can benefit from IPsec
tunnel-mode security, when NAT device acts as the IPsec tunnel end
point.

Section 2 below defines terms specific to this document.

Section 3 describes how tunnel mode IPsec security can be recognized
on NAT devices. IPsec Security architecture, format and operation of
various types of security mechanisms may be found in [Ref 2], [Ref 3]
and [Ref 4]. This section does not address how session keys and
policies are exchanged between a NAT device acting as IPsec gateway
and external peering nodes. The exchange could have taken place
manually or using any of known automatic exchange techniques.

Section 4 assumes that Public Key based IKE protocol [Ref 5] may be
used to automate exchange of security policies, session keys and
other Security Association (SA) attributes. This section describes a
method by which security policies administered for a private domain
may be translated for communicating with external nodes. Detailed
description of IKE protocol operation may be found in [Ref 5] and
[Ref 6].

Section 5 describes applications of the security model described in
the document. Applications listed include secure external realm
connectivity for private domain hosts and secure remote Access to
enterprise mobile hosts.

2. Terminology

Definitions for majority of terms used in this document may be found
in one of (a) NAT Terminology and Considerations document [Ref 1],
(b) IP security Architecture document [Ref 2], or (c) Internet Key
Enchange (IKE) document [Ref 5]. Below are terms defined specifically
for this document.

2.1. Normal-NAT

The term "Normal-NAT" is introduced to distinguish normal NAT
processing from the NAT processing used for secure packets embedded
within an IPsec secure tunnel. "Normal-NAT" is the normal NAT
processing as described in [Ref 1].

2.2. IPsec Policy Controlled NAT (IPC-NAT)

The term "IPsec Policy Controlled NAT" (IPC-NAT, for short) is
defined to describe the NAT transformation applied as an extension of
IPsec transformation to packets embedded within an IP-IP tunnel, for

which the NAT node is a tunnel end point. IPC-NAT function is
essentially an adaptation of NAT extensions to embedded packets of
tunnel-mode IPsec. Packets subject to IPC-NAT processing are
beneficiaries of IPsec security between the NAT device and an
external peer entity, be it a host or a gateway node.

IPsec policies place restrictions on what NAT mappings are used. For
example, IPsec access control security policies to a peer gateway
will likely restrict communication to only certain addresses and/or
port numbers. Thus, when NAT performs translations, it must insure
that the translations it performs are consist with the security
policies.

Just as with Normal-NAT, IPC-NAT function can assume any of NAT
flavors, including Traditional-NAT, Bi-directional-NAT and Twice-NAT.
An IPC-NAT device would support both IPC-NAT and normal-NAT
functions.

3. Security model of IPC-NAT

The IP security architecture document [Ref 2] describes how IP
network level security may be accomplished within a globally unique
address realm. Transport and tunnel mode security are discussed. For
purposes of this document, we will assume IPsec security to mean
tunnel mode IPsec security, unless specified otherwise. Elements
fundamental to this security architecture are (a) Security Policies,
that determine which packets are permitted to be subject to Security
processing, and (b) Security Association Attributes that identify the
parameters for security processing, including IPsec protocols,
algorithms and session keys to be applied.

Operation of tunnel mode IPsec security on a device that does not
support Network Address Translation may be described as below in
figures 1 and 2.

+---------------+ No +---------------------------+
+--->Forward packet in the Clear
Outgoing Does the packet Or Drop, as appropriate.
-------->match Outbound - +---------------------------+
Packet Security +-------------+
Policies? Yes Perform Forward
+--->Outbound --------->
+---------------+ Security IPsec Pkt
(Tunnel Mode)
+-------------+

Figure 1. Operation of Tunnel-Mode IPsec on outgoing packets.

IPsec packet +----------+ +----------+
destined to Perform Embedded Does the No(Drop)
------------>Inbound --------->Pkt match -------->
the device Security Packet Inbound SA Yes(Forward)
(Detunnel) Policies?
+----------+ +----------+

Figure 2. Operation of Tunnel-Mode IPsec on Incoming packets

A NAT device that provides tunnel-mode IPsec security would be
required to administer security policies based on private realm
addressing. Further, the security policies determine the IPsec tunnel
end-point peer. As a result, a packet may be required to undergo
different type of NAT translation depending upon the tunnel end-point
the IPsec node peers with. In other Words, IPC-NAT will need a unique
set of NAT maps for each security policy configured. IPC-NAT will
perform address translation in conjunction with IPsec processing
differently with each peer, based on security policies. The
following diagrams (figure 3 and figure 4) illustrate the operation
of IPsec tunneling in conjunction with NAT. Operation of an IPC-NAT
device may be distinguished from that of an IPsec gateway that does
not support NAT as follows.

(1) IPC-NAT device has security policies administered using
private realm addressing. A traditional IPsec gateway will
have its security policies administered using a single realm
(say, external realm) addressing.

(2) Elements fundamental to the security model of an IPC-NAT
device includes IPC-NAT address mapping (and other NAT
parameter definitions) in conjunction with Security policies
and SA attributes. Fundamental elements of a traditional
IPsec gateway are limited only to Security policies and SA
attributes.

+---------------+ +-------------------------+
No Apply Normal-NAT or Drop
Outgoing Does the packet +---> as appropriate
-------->match Outbound - +-------------------------+
Packet Security +---------+ +-------------+
(Private Policies? Yes Perform Perform Forward
Domain) +--->Outbound ->Outbound -------->
+---------------+ NAT Security IPsec Pkt
(IPC-NAT) (Tunnel mode)
+---------+ +-------------+

Figure 3. Tunnel-Mode IPsec on an IPC-NAT device for outgoing pkts

IPsec Pkt +----------+ +---------+ +----------+
destined Perform Embedded Perform Does the No(Drop)
--------->Inbound --------->Inbound ->Pkt match -------->
to device Security Packet NAT Inbound SAYes(Forward)
(External (Detunnel) (IPC-NAT) Policies?
Domain) +----------+ +---------+ +----------+

Figure 4. Tunnel-Mode IPsec on an IPC-NAT device for Incoming pkts

Traditional NAT is session oriented, allowing outbound-only sessions
to be translated. All other flavors of NAT are Bi-directional. Any
and all flavors of NAT mapping may be used in conjunction with the
security policies and secure processing on an IPC-NAT device. For
illustration purposes in this document, we will assume tunnel mode
IPsec on a Bi-directional NAT device.

Notice however that a NAT device capable of providing security across
IPsec tunnels can continue to support Normal-NAT for packets that do
not require IPC-NAT. Address mapping and other NAT parameter
definitions for Normal-NAT and IPC-NAT are distinct. Figure 3
identifies how a NAT device distinguishes between outgoing packets
that need to be processed through Normal-NAT vs. IPC-NAT. As for
packets incoming from external realm, figure 4 outlines packets that
may be subject to IPC-NAT. All other packets are subject to Normal-
NAT processing only.

4. Operation of IKE protocol on IPC-NAT device.

IPC-NAT operation described in the previous section can be
accomplished based on manual session key exchange or using an
automated key Exchange protocol between peering entities. In this
section, we will consider adapting IETF recommended Internet Key
Exchange (IKE) protocol on a IPC-NAT device for automatic exchange of
security policies and SA parameters. In other words, we will focus on
the operation of IKE in conjunction with tunnel mode IPsec on NAT
devices. For the reminder of this section, we will refer NAT device
to mean IPC-NAT device, unless specified otherwise.

IKE is based on UDP protocol and uses public-key encryption to
exchange session keys and other attributes securely across an address
realm. The detailed protocol and operation of IKE in the context of
IP may be found in [Ref 3] and [Ref 4]. Essentially, IKE has 2
phases.

In the first phase, IKE peers operate in main or aggressive mode to
authenticate each other and set up a secure channel between
themselves. A NAT device has an interface to the external realm and
is no different from any other node in the realm to negotiate phase I

with peer external nodes. The NAT device may assume any of the valid
Identity types and authentication methodologies necessary to
communicate and authenticate with peers in the realm. The NAT device
may also interface with a Certification Authority (CA) in the realm
to retrieve certificates and perform signature validation.

In the second phase, IKE peers operate in Quick Mode to exchange
policies and IPsec security proposals to negotiate and agree upon
security transformation algorithms, policies, keys, lifetime and
other security attributes. During this phase, IKE process must
communicate with IPsec Engine to (a) collect secure session
attributes and other parameters to negotiate with peer IKE nodes,
and to (b) notify security parameters agreed upon (with peer) during
the negotiation.

An IPC-NAT device, operating as IPsec gateway, has the security
policies administered based on private realm addressing. An ALG will
be required to translate policies from private realm addressing into
external addressing, as the IKE process needs to communicate these
policies to peers in external realm. Note, IKE datagrams are not
subject to any NAT processing. IKE-ALG simply translates select
portions of IKE payload as per the NAT map defined for the policy
match. The following diagram illustrates how an IKE-ALG process
interfaces with IPC-NAT to take the security policies and IPC-NAT
maps and generates security policies that IKE could communicate
during quick mode to peers in the external realm.

Policies in quick mode are exchanged with a peer as a combination of
IDci and IDcr payloads. The combination of IDs (policies) exchanged
by each peer must match in order for the SA parameters on either end
to be applied uniformly. If the IDs are not exchanged, the assumption
would be that the Quick mode negotiated SA parameters are applicable
between the IP addresses assumed by the main mode.

Depending on the nature of security policies in place(ex: end-to-end
sessions between a pair of nodes vs. sessions with an address range),
IKE-ALG may need to request NAT to set up address bindings and/or
transport bindings for the lifetime (in seconds or Kilo-Bytes) the
sessions are negotiated. In the case the ALG is unable to setup the
necessary address bindings or transport bindings, IKE-ALG will not be
able to translate security policies and that will result in IKE not
pursuing phase II negotiation for the effected policies.

When the Negotiation is complete and successful, IKE will communicate
the negotiated security parameters directly to the IPC-NAT gateway
engine as described in the following diagram.

+---------+

Negotiated Security Parameters IKE
+-------------------------------- Process
(including session Keys)
+---------+
^ ^
Translated
Secure Security
Policies Proposals
v
+---------+ Security Policies, based +---------+
------------------------->
on Pvt. realm addressing
IPC-NAT
(IPsec IPC-NAT MAPs IKE-ALG
Gateway)------------------------->

Security Proposals
------------------------->

NAT Control exchange
<------------------------>
+---------+ +---------+

Figure 5. IKE-ALG translates Security policies, using NAT Maps.

5. Applications of IPC-NAT security model

IPC-NAT operational model described thus far illustrates how a NAT
device can be used as an IPsec tunnel end point to provide secure
transfer of data in external realm. This section will attempt to
illustrate two applications of such a model.

5.1. Secure Extranet Connectivity

IPC-NAT Model has a direct application of being able to provide clear
as well as secure connectivity with external realm using a NAT
device. In particular, IPC-NAT device at the border of a private
realm can peer with an IPsec gateway of an external domain to secure
the Extranet connection. Extranet refers to the portion of the path
that crosses the Internet between peering gateway nodes.

5.2. Secure Remote Access to Mobile Users of an Enterprise

Say, a node from an enterprise moves out of the enterprise, and
attempts to connect to the enterprise from remote site, using a
temporary service provider assigned address (Care-of-Address). In
such a case, the mobile user could setup an IPsec tunnel session with
the corporate IPC-NAT device using a user-ID and authentication
mechanism that is agreed upon. Further, the user may be configured
with enterprise DNS server, as an extension of authentication
following IKE Phase I. This would allow the user to access enterprise
resources by name.

However, many enterprise servers and applications rely on source IP
address for authentication and deny access for packets that do not
originate from the enterprise address space. In these scenarios,
IPC-NAT has the ability (unlike a traditional IPsec gateway) to
perform Network Address Translation (NAT) for remote access users, so
their temporary address in external realm is translated into a
enterprise domain address, while the packets are within private
realm. The flavor of IPC-NAT performed would be traditional NAT
(i.e., assuming mobile-user address space to be private realm and
Enterprise address space to be external realm), which can either be
Basic NAT (using a block of enterprise addresses for translation) or
NAPT(using a single enterprise address for translation).

The secure remote access application described is pertinent to all
enterprises, irrespective of whether an enterprise uses IANA
registered addresses or not.

The secure remote access application described is different from
Mobile-IP in that, the mobile node (described in this application)
does not retain the Home-Network address and simply uses the Care-
Of-address for communication purposes. It is conceivable for the
IPC-NAT Gateway to transparently provide Mobile-IP type connectivity
to the Mobile node by binding the mobile node's Care-of-Address with
its Home Address. Provision of such an address mapping to IPC-NAT
gateway, however, is not within the scope of this document.

6. Security Considerations

If NATs and ALGs are not in a trusted boundary, that is a major
security problem, as ALGs snoop end user traffic payload.
Application level payload could be encrypted end-to-end, so long as
the payload does not contain IP addresses and/or transport
identifiers that are valid in only one of the realms. With the
exception of Realm-Specific IP, end-to-end IP network level security
assured by current IPsec techniques is not attainable with NATs in
between. The IPC-NAT model described in this document outlines an

approach by which network level security may be oBTained within
external realm.

NATs, when combined with ALGs, can ensure that the datagrams injected
into Internet have no private addresses in headers or payload.
Applications that do not meet these requirements may be dropped using
firewall filters. For this reason, it is not uncommon to find that
IPC-NATs, ALGs and firewalls co-exist to provide security at the
border of a private network.

REFERENCES

[1] Srisuresh, P. and M. Holdrege, "IP Network Address Translator
(NAT) Terminology and Considerations", RFC2663, August 1999.

[2] Kent, S. and R. Atkinson, "Security Architecture for the
Internet Protocol", RFC2401, November 1998

[3] Kent, S. and R. Atkinson, "IP Encapsulating Security Payload
(ESP)", RFC2406, November 1998

[4] Kent, S. and R. Atkinson, "IP Authentication Header", RFC2402,
November 1998.

[5] Harkins, D. and D. Carrel, "The Internet Key Exchange (IKE)",
RFC2409, November 1998.

[6] Piper, D., "The Internet IP Security Domain of Interpretation
for ISAKMP", RFC2407, November 1998.

[7] Carpenter, B., Crowcroft, J. and Y. Rekhter, "IPv4 Address
Behavior Today", RFC2101, February 1997.

[8] Rekhter, Y., Moskowitz, B., Karrenberg, D., de Groot G. and E.
Lear, "Address Allocation for Private Internets", BCP 5, RFC
1918, February 1996.

Author's Address

Pyda Srisuresh
Lucent technologies
4464 Willow Road
Pleasanton, CA 94588-8519
U.S.A.

Phone: (925) 737-2153
Fax: (925) 737-2110
EMail: srisuresh@lucent.com

Full Copyright Statement

Copyright (C) The Internet Society (1999). 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 and derivative works. However, this
document itself may not be modified in any way, such as by removing
the copyright notice or references to the Internet Society or other
Internet organizations, except as needed for the purpose of
developing Internet standards in which case the procedures for
copyrights defined in the Internet Standards process must be
followed, or as required to translate it into languages other than
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.

Acknowledgement

Funding for the RFCEditor function is currently provided by the
Internet Society.