Network Working Group                                        N. Williams
Request for Comments: 5386                                           Sun
Category: Standards Track                                  M. Richardson
                                                                    SSW
                                                          November 2008


    Better-Than-Nothing Security: An Unauthenticated Mode of IPsec

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) 2008 IETF Trust and the persons identified as the
  document authors.  All rights reserved.

  This document is subject to BCP 78 and the IETF Trust's Legal
  Provisions Relating to IETF Documents
  (http://trustee.ietf.org/license-info) in effect on the date of
  publication of this document.  Please review these documents
  carefully, as they describe your rights and restrictions with respect
  to this document.

Abstract

  This document specifies how to use the Internet Key Exchange (IKE)
  protocols, such as IKEv1 and IKEv2, to setup "unauthenticated"
  security associations (SAs) for use with the IPsec Encapsulating
  Security Payload (ESP) and the IPsec Authentication Header (AH).  No
  changes to IKEv2 bits-on-the-wire are required, but Peer
  Authorization Database (PAD) and Security Policy Database (SPD)
  extensions are specified.  Unauthenticated IPsec is herein referred
  to by its popular acronym, "BTNS" (Better-Than-Nothing Security).












Williams & Richardson       Standards Track                     [Page 1]

RFC 5386                       BTNS IPsec                  November 2008


Table of Contents

  1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  2
    1.1.  Conventions Used in This Document  . . . . . . . . . . . .  2
  2.  BTNS . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  3
  3.  Usage Scenarios  . . . . . . . . . . . . . . . . . . . . . . .  5
    3.1.  Example #1: A Security Gateway . . . . . . . . . . . . . .  5
    3.2.  Example #2: A Mixed End-System . . . . . . . . . . . . . .  7
    3.3.  Example #3: A BTNS-Only System . . . . . . . . . . . . . .  8
    3.4.  Miscellaneous Comments . . . . . . . . . . . . . . . . . .  9
  4.  Security Considerations  . . . . . . . . . . . . . . . . . . .  9
    4.1.  Connection Latching and Channel Binding  . . . . . . . . .  9
    4.2.  Leap-of-Faith (LoF) for BTNS . . . . . . . . . . . . . . . 10
  5.  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 10
  6.  References . . . . . . . . . . . . . . . . . . . . . . . . . . 10
    6.1.  Normative References . . . . . . . . . . . . . . . . . . . 10
    6.2.  Informative References . . . . . . . . . . . . . . . . . . 10

1.  Introduction

  Here we describe how to establish unauthenticated IPsec SAs using
  IKEv2 [RFC4306] and unauthenticated public keys.  No new on-the-wire
  protocol elements are added to IKEv2.

  The [RFC4301] processing model is assumed.

  This document does not define an opportunistic BTNS mode of IPsec
  whereby nodes may fall back to unprotected IP when their peers do not
  support IKEv2, nor does it describe "leap-of-faith" modes or
  "connection latching".

  See [RFC5387] for the applicability and uses of BTNS and definitions
  of these terms.

  This document describes BTNS in terms of IKEv2 and [RFC4301]'s
  concepts.  There is no reason why the same methods cannot be used
  with IKEv1 [RFC2408], [RFC2409], and [RFC2401]; however, those
  specifications do not include the PAD concepts, and therefore it may
  not be possible to implement BTNS on all compliant RFC2401
  implementations.

1.1.  Conventions Used in This Document

  The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
  "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
  document are to be interpreted as described in [RFC2119].





Williams & Richardson       Standards Track                     [Page 2]

RFC 5386                       BTNS IPsec                  November 2008


2.  BTNS

  The IPsec processing model is hereby modified as follows:

  o  A new ID type is added: 'PUBLICKEY'.  IDs of this type have public
     keys as values.  This ID type is not used on the wire.

  o  PAD entries that match on PUBLICKEY IDs are referred to as "BTNS
     PAD entries".  All other PAD entries are referred to as "non-BTNS
     PAD entries".

  o  BTNS PAD entries may match on specific peer PUBLICKEY IDs (or
     public key fingerprints) or on all peer public keys.  The latter
     is referred to as the "wildcard BTNS PAD entry".

  o  BTNS PAD entries MUST logically (see below) follow all other PAD
     entries (the PAD being an ordered list).

  o  At most one wildcard BTNS PAD entry may appear in the PAD, and, if
     present, MUST be the last entry in the PAD (see below).

  o  Any peer that uses an IKEv2 AUTH method involving a digital
     signature (made with a private key to a public key cryptosystem)
     may match a BTNS PAD entry, provided that it matches no non-BTNS
     PAD entries.  Suitable AUTH methods as of August 2007 are: RSA
     Digital Signature (method #1) and DSS Digital Signature (method
     #3); see [RFC4306], Section 3.8.

  o  A BTNS-capable implementation of IPsec will first search the PAD
     for non-BTNS entries matching a peer's ID.  If no matching
     non-BTNS PAD entries are found, then the peer's ID MUST be coerced
     to be of 'PUBLICKEY' type with the peer's public key as its value.
     The PAD is then searched again for matching BTNS PAD entries.
     This ensures that BTNS PAD entries logically follow non-BTNS PAD
     entries.  A single PAD search that preserves these semantics is
     allowed.

  o  A peer that matches a BTNS PAD entry is referred to as a "BTNS
     peer".  Such a peer is "authenticated" by verifying the signature
     in its IKEv2 AUTH payload with the public key from the peer's CERT
     payload.

  o  Of course, if no matching PAD entry is found, then the IKE SA is
     rejected as usual.







Williams & Richardson       Standards Track                     [Page 3]

RFC 5386                       BTNS IPsec                  November 2008


  o  A new flag for SPD entries: 'BTNS_OK'.  Traffic to/from peers that
     match the BTNS PAD entry will match only SPD entries that have the
     BTNS_OK flag set.  The SPD may be searched by address or by ID (of
     type PUBLICKEY for BTNS peers), as per the IPsec processing model
     [RFC4301].  Searching by ID in this case requires creation of SPD
     entries that are bound to public key values.  This could be used
     to build "leap-of-faith" [RFC5387] behavior (see Section 4.2), for
     example.

  Nodes MUST reject IKE_SA proposals from peers that match non-BTNS PAD
  entries but fail to authenticate properly.

  Nodes wishing to be treated as BTNS nodes by their peers MUST include
  bare public key CERT payloads.  Currently only bare RSA public key
  CERT payloads are defined, which means that BTNS works only with RSA
  public keys at this time (see "Raw RSA Key" in Section 3.6 of
  [RFC4306]).  Nodes MAY also include any number of certificates that
  bind the same public key.  These certificates do not need to be
  pre-shared with their peers (e.g., because ephemeral, self-signed).
  RSA keys for use in BTNS may be generated at any time, but connection
  latching [ConnLatch] requires that they remain constant between IKEv2
  exchanges that are used to establish SAs for latched connections.

  To preserve standard IPsec access control semantics:

  o  BTNS PAD entries MUST logically follow all non-BTNS PAD entries,

  o  the wildcard BTNS PAD entry MUST be the last entry in the PAD,
     logically, and

  o  the wildcard BTNS PAD entry MUST have ID constraints that do not
     logically overlap those of other PAD entries.

  As described above, the logical PAD ordering requirements can easily
  be implemented by searching the PAD twice at peer authentication
  time: once using the peer-asserted ID, and if that fails, once using
  the peer's public key as a PUBLICKEY ID.  A single pass
  implementation that meets this requirement is permitted.

  The BTNS entry ID constraint non-overlap requirement can easily be
  implemented by searching the PAD twice: once when BTNS peers
  authenticate, and again when BTNS peers negotiate child SAs.  In the
  first pass, the PAD is searched for a matching PAD entry as described
  above.  In the second, it is searched to make sure that BTNS peers'
  asserted child SA traffic selectors do not conflict with non-BTNS PAD
  entries.  Single pass implementations that preserve these semantics
  are feasible.




Williams & Richardson       Standards Track                     [Page 4]

RFC 5386                       BTNS IPsec                  November 2008


3.  Usage Scenarios

  In order to explain the above rules, a number of scenarios will be
  examined.  The goal here is to persuade the reader that the above
  rules are both sufficient and necessary.

  This section is informative only.

  To explain the scenarios, a reference diagram describing an example
  network will be used.  It is as follows:

                            [Q]  [R]
       AS1                   .    .              AS2
    [A]----+----[SG-A].......+....+.......[SG-B]-------[B]
                             ......               \
                             ..PI..                ----[btns-B]
                             ......
                [btns-C].....+....+.......[btns-D]

                   Figure 1: Reference Network Diagram

  In this diagram, there are eight systems.  Six systems are end-nodes
  (A, B, C, D, Q, and R).  Two are security gateways (SG-A, SG-B)
  protecting networks on which [A] and [B] reside.  Node [Q] is IPsec
  and BTNS capable.  Node [R] is a simple node, with no IPsec or BTNS
  capability.  Nodes [C] and [D] are BTNS capable.

  Nodes [C] and [Q] have fixed addresses.  Node [D] has a non-fixed
  address.

  We will examine how these various nodes communicate with node [SG-A]
  and/or how [SG-A] rejects communications with some such nodes.  In
  the first example, we examine [SG-A]'s point of view.  In the second
  example, we look at [Q]'s point of view.  In the third example, we
  look at [C]'s point of view.

  PI is the Public Internet ("The Wild").

3.1.  Example #1: A Security Gateway

  The machine that we will focus on in this example is [SG-A], a
  firewall device of some kind that we wish to configure to respond to
  BTNS connections from [C].








Williams & Richardson       Standards Track                     [Page 5]

RFC 5386                       BTNS IPsec                  November 2008


  [SG-A] has the following PAD and SPD entries:

                               Child SA
        Rule Remote ID        IDs allowed  SPD Search by
        ---- ---------        -----------  -------------
         1   <B's ID>         <B's network>  by-IP
         2   <Q's ID>         <Q's host>     by-IP
         3   PUBLICKEY:any         ANY       by-IP

  The last entry is the BTNS entry.

                       Figure 2: [SG-A] PAD Table

  Note that [SG-A]'s PAD entry has one and only one wildcard PAD entry:
  the BTNS catch-all PAD entry as the last entry, as described in
  Section 2.

  <Child SA IDs allowed> and <SPD Search by> are from [RFC4301],
  Section 4.4.3.

        Rule Local Remote Next Layer BTNS  Action
              addr  addr   Protocol   ok
        ---- ----- ------ ---------- ----  -----------------------
         1   [A]    [R]      ANY      N/A  BYPASS
         2   [A]    [Q]      ANY      no   PROTECT(ESP,tunnel,AES,
                                                       SHA256)
         3   [A]     B-net   ANY      no   PROTECT(ESP,tunnel,AES,
                                                       SHA256)
         4   [A]     ANY     ANY      yes  PROTECT(ESP,transport,
                                                       integr+conf)

                       Figure 3: [SG-A] SPD Table

  The processing by [SG-A] of SA establishment attempts by various
  peers is as follows:

  o  [Q] does not match PAD entry #1 but does match PAD entry #2.  PAD
     processing stops, then the SPD is searched by [Q]'s ID to find
     entry #2.  CHILD SAs are then allowed that have [SG-A]'s and [Q]'s
     addresses as the end-point addresses.

  o  [SG-B] matches PAD entry #1.  PAD processing stops, then the SPD
     is searched by [SG-B]'s ID to find SPD entry #3.  CHILD SAs are
     then allowed that have [SG-A]'s address and any addresses from B's
     network as the end-point addresses.

  o  [R] does not initiate any IKE SAs; its traffic to [A] is bypassed
     by SPD entry #1.



Williams & Richardson       Standards Track                     [Page 6]

RFC 5386                       BTNS IPsec                  November 2008


  o  [C] does not match PAD entries #1 or #2 but does match entry #3,
     the BTNS wildcard PAD entry.  The SPD is searched by [C]'s
     address, and SPD entry #4 is matched.  CHILD SAs are then allowed
     that have [SG-A]'s address and [C]'s address as the end-point
     addresses, provided that [C]'s address is neither [Q]'s nor any of
     [B]'s (see Section 2).  See the last bullet item below.

  o  A rogue BTNS node attempting to assert [Q]'s or [B]'s addresses
     will either match the PAD entries for [Q] or [B] and fail to
     authenticate as [Q] or [B], in which case they are rejected, or
     they will match PAD entry #3 but will not be allowed to create
     CHILD SAs with [Q]'s or [B]'s addresses as traffic selectors.

  o  A rogue BTNS node attempting to establish an SA whereby the rogue
     node asserts [C]'s address will succeed at establishing such an
     SA.  Protection for [C] requires additional bindings of [C]'s
     specific BTNS ID (that is, its public key) to its traffic flows
     through connection latching and channel binding or through leap-
     of-faith, none of which are described here.

3.2.  Example #2: A Mixed End-System

  [Q] is an NFSv4 server.

  [Q] is a native IPsec implementation, and its NFSv4 implementation is
  IPsec-aware.

  [Q] wants to protect all traffic with [A].  [Q] also wants to protect
  NFSv4 traffic with all peers.  Its PAD and SPD are configured as
  follows:

                               Child SA
        Rule Remote ID        IDs allowed  SPD Search by
        ---- ---------        -----------  -------------
         1   <[A]'s ID>       <[A]'s address> by-IP
         2   PUBLICKEY:any    ANY             by-IP

  The last entry is the BTNS entry.

                         Figure 4: [Q] PAD Table











Williams & Richardson       Standards Track                     [Page 7]

RFC 5386                       BTNS IPsec                  November 2008


        Rule Local Remote Next Layer BTNS  Action
              addr  addr   Protocol   ok
        ---- ----- ------ ---------- ----  -----------------------
         1    [Q]    [A]     ANY      no   PROTECT(ESP,tunnel,AES,
                                                       SHA256)
         2    [Q]    ANY     ANY      yes  PROTECT(ESP,transport,
              with                                      integr+conf)
            port 2049

                         Figure 5: [Q] SPD Table

  The same analysis shown above in Section 3.1 applies here with
  respect to [SG-A], [C], and rogue peers.  The second SPD entry
  permits any BTNS-capable node to negotiate a port-specific SA to port
  2049, the port on which NFSv4 runs.  Additionally, [SG-B] is treated
  as a BTNS peer as it is not known to [Q], and therefore any host
  behind [SG-B] can access the NFSv4 service on [Q].  As [Q] has no
  formal relationship with [SG-B], rogues can impersonate [B] (i.e.,
  assert [B]'s addresses).

3.3.  Example #3: A BTNS-Only System

  [C] supports only BTNS and wants to use BTNS to protect NFSv4
  traffic.  Its PAD and SPD are configured as follows:

                               Child SA
        Rule Remote ID        IDs allowed  SPD Search by
        ---- ---------        -----------  -------------
         1   PUBLICKEY:any    ANY          by-IP

  The last (and only) entry is the BTNS entry.

                         Figure 6: [Q] PAD Table


        Rule Local Remote Next Layer BTNS  Action
              addr  addr   Protocol   ok
        ---- ----- ------ ---------- ----  -----------------------
         1    [C]    ANY      ANY      yes  PROTECT(ESP,transport,
                    with                               integr+conf)
                    port
                    2049

         2    [C]    ANY      ANY      N/A  BYPASS

                       Figure 7: [SG-A] SPD Table





Williams & Richardson       Standards Track                     [Page 8]

RFC 5386                       BTNS IPsec                  November 2008


  The analysis from Section 3.1 applies as follows:

  o  Communication with [Q] on port 2049 matches SPD entry number 1.
     This causes [C] to initiate an IKEv2 exchange with [Q].  The PAD
     entry on [C] causes it to not care what identity [Q] asserts.
     Further authentication (and channel binding) could occur within
     the NFSv4 protocol.

  o  Communication with [A], [B], or any other internet machine
     (including [Q]), occurs in the clear, so long as it is not on port
     2049.

  o  All analysis about rogue BTNS nodes applies, but they can only
     assert SAs for port 2049.

3.4.  Miscellaneous Comments

  If [SG-A] were not BTNS capable, then it would not have PAD and SPD
  entries #3 and #4, respectively, in example #1.  Then [C] would be
  rejected as usual under the standard IPsec model [RFC4301].

  Similarly, if [Q] were not BTNS capable, then it would not have PAD
  and SPD entries #2 in example #2.  Then [C] would be rejected as
  usual under the standard IPsec model [RFC4301].

4.  Security Considerations

  Unauthenticated security association negotiation is subject to man-
  in-the-middle (MITM) attacks and should be used with care.  Where
  security infrastructures are lacking, this may indeed be better than
  nothing.

  Use with applications that bind authentication at higher network
  layers to secure channels at lower layers may provide one secure way
  to use unauthenticated IPsec, but this is not specified herein.

  The BTNS PAD entry must be last and its child SA ID constraints must
  be non-overlapping with any other PAD entry, as described in
  Section 2.  This will ensure that no BTNS peer can impersonate
  another IPsec non-BTNS peer.

4.1.  Connection Latching and Channel Binding

  BTNS is subject to MITM attacks.  One way to protect against MITM
  attacks subsequent to initial communications is to use "connection
  latching" [ConnLatch].  In connection latching, upper layer protocols
  (ULPs) cooperate with IPsec to bind discrete packet flows to




Williams & Richardson       Standards Track                     [Page 9]

RFC 5386                       BTNS IPsec                  November 2008


  sequences of similar SAs.  Connection latching requires native IPsec
  implementations.

  MITMs can be detected by using application-layer authentication
  frameworks and/or mechanisms, such as the GSS-API [RFC2743], with
  channel binding [RFC5056].  IPsec "channels" are nothing other than
  latched connections.

4.2.  Leap-of-Faith (LoF) for BTNS

  "Leap of faith" is the term generally used when a user accepts the
  assertion that a given key identifies a peer on the first
  communication (despite a lack of strong evidence for that assertion),
  and then remembers this association for future communications.
  Specifically this is a common mode of operation for Secure Shell
  [RFC4251] clients.  When a server is encountered for the first time,
  the Secure Shell client may ask the user whether to accept the
  server's public key.  If so, the client records the server's name (as
  given by the user) and public key in a database.

  Leap of faith can work in a similar way for BTNS nodes, but it is
  currently still being designed and specified by the IETF BTNS WG.

5.  Acknowledgements

  Thanks to the following reviewer: Stephen Kent.

6.  References

6.1.  Normative References

  [RFC2119]    Bradner, S., "Key words for use in RFCs to Indicate
               Requirement Levels", BCP 14, RFC 2119, March 1997.

  [RFC4301]    Kent, S. and K. Seo, "Security Architecture for the
               Internet Protocol", RFC 4301, December 2005.

6.2.  Informative References

  [ConnLatch]  Williams, N., "IPsec Channels: Connection Latching",
               Work in Progress, April 2008.

  [RFC2401]    Kent, S. and R. Atkinson, "Security Architecture for the
               Internet Protocol", RFC 2401, November 1998.

  [RFC2408]    Maughan, D., Schneider, M., and M. Schertler, "Internet
               Security Association and Key Management Protocol
               (ISAKMP)", RFC 2408, November 1998.



Williams & Richardson       Standards Track                    [Page 10]

RFC 5386                       BTNS IPsec                  November 2008


  [RFC2409]    Harkins, D. and D. Carrel, "The Internet Key Exchange
               (IKE)", RFC 2409, November 1998.

  [RFC2743]    Linn, J., "Generic Security Service Application Program
               Interface Version 2, Update 1", RFC 2743, January 2000.

  [RFC4251]    Ylonen, T. and C. Lonvick, "The Secure Shell (SSH)
               Protocol Architecture", RFC 4251, January 2006.

  [RFC4306]    Kaufman, C., "Internet Key Exchange (IKEv2) Protocol",
               RFC 4306, December 2005.

  [RFC5056]    Williams, N., "On the Use of Channel Bindings to Secure
               Channels", RFC 5056, November 2007.

  [RFC5387]    Touch, J., Black, D., and Y. Wang, "Problem and
               Applicability Statement for Better-Than-Nothing Security
               (BTNS)", RFC 5387, November 2008.

Authors' Addresses

  Nicolas Williams
  Sun Microsystems
  5300 Riata Trace Ct
  Austin, TX  78727
  US

  EMail: [email protected]


  Michael C. Richardson
  Sandelman Software Works
  470 Dawson Avenue
  Ottawa, ON  K1Z 5V7
  CA

  EMail: [email protected]
  URI:   http://www.sandelman.ottawa.on.ca/













Williams & Richardson       Standards Track                    [Page 11]