Network Working Group                                        S. Bellovin
Request for Comments: 1675                        AT&T Bell Laboratories
Category: Informational                                      August 1994


                      Security Concerns for IPng

Status of this Memo

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

Abstract

  This document was submitted to the IETF IPng area in response to RFC
  1550.  Publication of this document does not imply acceptance by the
  IPng area of any ideas expressed within.  Comments should be
  submitted to the [email protected] mailing list.

Overview and Rationale

  A number of the candidates for IPng have some features that are
  somewhat worrisome from a security perspective.  While it is not
  necessary that IPng be an improvement over IPv4, it is mandatory that
  it not make things worse.  Below, I outline a number of areas of
  concern.  In some cases, there are features that would have a
  negative impact on security if nothing else is done.  It may be
  desirable to adopt the features anyway, but in that case, the
  corrective action is mandatory.

Firewalls

  For better or worse, firewalls are very much a feature of today's
  Internet.  They are not, primarily, a response to network protocol
  security problems per se.  Rather, they are a means to compensate for
  failings in software engineering and system administration.  As such,
  firewalls are not likely to go away any time soon; IPng will do
  nothing to make host programs any less buggy.  Anything that makes
  firewalls harder to deploy will make IPng less acceptable in the
  market.

  Firewalls impose a number of requirements.  First, there must be a
  hierarchical address space.  Many address-based filters use the
  structure of IPv4 addresses for access control decisions.
  Fortunately, this is a requirement for scalable routing as well.





Bellovin                                                        [Page 1]

RFC 1675               Security Concerns for IPng            August 1994


  Routers, though, only need access to the destination address of the
  packet.  Network-level firewalls often need to check both the source
  and destination address.  A structure that makes it harder to find
  the source address is a distinct negative.

  There is also a need for access to the transport-level (i.e., the TCP
  or UDP) header.  This may be for the port number field, or for access
  to various flag bits, notably the ACK bit in the TCP header.  This
  latter field is used to distinguish between incoming and outgoing
  calls.

  In a different vein, at least one of the possible transition plans
  uses network-level packet translators [1].  Organizations that use
  firewalls will need to deploy their own translators to aid in
  converting their own internal networks.  They cannot rely on
  centrally-located translators intended to serve the entire Internet
  community.  It is thus vital that translators be simple, portable to
  many common platforms, and cheap -- we do not want to impose too high
  a financial barrier for converts to IPng.

  By the same token, it is desirable that such translation boxes not be
  usable for network-layer connection-laundering.  It is difficult
  enough to trace back attacks today; we should not make it harder.
  (Some brands of terminal servers can be used for laundering.  Most
  sites with such boxes have learned to configure them so that such
  activities are impossible.)  Comprehensive logging is a possible
  alternative.

  IPAE [1] does not have problems with its translation strategy, as
  address are (insofar as possible) preserved; it is necessary to avoid
  any alternative strategies, such as circuit-level translators, that
  might.

Encryption and Authentication

  A number of people are starting to experiment with IP-level
  encryption and cryptographic authentication.  This trend will (and
  should) continue.  IPng should not make this harder, either
  intrinsically or by imposing a substantial perforance barrier.

  Encryption can be done with various different granularities: host to
  host, host to gateway, and gateway to gateway.  All of these have
  their uses; IPng must not rule out any of them.  Encapsulation and
  tunneling strategies are somewhat problematic, as the packet may no
  longer carry the original source address when it reaches an
  encrypting gateway.  (This may be seen more as a constraint on
  network topologies.  So be it, but we should warn people of the
  limitation.)



Bellovin                                                        [Page 2]

RFC 1675               Security Concerns for IPng            August 1994


  Dual-stack approaches, such as in TUBA's transition plan [2], imply
  multiple addresses for each host.  (IPAE has this feature, too.) The
  encryption and access control infrastructure needs to know about all
  addresses for a given host, belonging to whichever stack.  It should
  not be possible to bypass authentication or encryption by asking for
  a different address for the same host.

Source Routing and Address-based Authentication

  The dominant form of host authentication in today's Internet is
  address-based.  That is, hosts often decide to trust other hosts
  based on their IP addresses.  (Actually, it's worse than that; much
  authentication is name-based, which opens up new avenues of attack.
  But if an attacker can spoof an IP address, there's no need to attack
  the name service.)  To the extent that it does work, address-based
  authentication relies on the implied accuracy of the return route.
  That is, though it is easy to inject packets with a false source
  address, replies will generally follow the usual routing patterns,
  and be sent to the real host with that address.  This frustrates
  most, though not all, attempts at impersonation.

  Problems can arise if source-routing is used.  A source route, which
  must be reversed for reply packets, overrides the usual routing
  mechanism, and hence destroys the security of address-based
  authentication.  For this reason, many organizations disable source-
  routing, at least at their border routers.

  One candidate IPng -- SIPP -- includes source-routing as an important
  component.  To the extent this is used, it is a breaks address-based
  authentication.  This may not be bad; in fact, it is probably good.
  But it is vital that a more secure cryptographic authentication
  protocol be defined and deployed before any substantial cutover to
  source routing, if SIPP is adopted.

Accounting

  An significant part of the world wishes to do usage-sensitive
  accounting.  This may be for billing, or it may simply be to
  accomodate quality-of-service requests.  Either way, definitive
  knowledge of the relevant address fields is needed.  To accomodate
  this, IPng should have a non-intrusive packet authentication
  mechanism.  By "non-intrusive", I mean that it should (a) present
  little or no load to intermediate hops that do not need to do
  authentication; (b) be deletable (if desired) by the border gateways,
  and (c) be ignorable by end-systems or billing systems to which it is
  not relevant.





Bellovin                                                        [Page 3]

RFC 1675               Security Concerns for IPng            August 1994


References

  [1] Gilligan, R., and E. Nordmark, "IPAE: The SIPP Interoperability
      and Transition Mechanism", Work in Progress, March 16, 1994.

  [2] Piscitello, D., "Transition Plan for TUBA/CLNP", Work in
      Progress, March 4, 1994.

Security Consierations

  This entire memo is about Security Considerations.

Author's Address

  Steven M. Bellovin
  Software Engineering Research Department
  AT&T Bell Laboratories
  600 Mountain Avenue
  Murray Hill, NJ  07974, USA

  Phone: +1 908-582-5886
  Fax: +1 908-582-3063
  EMail:  [email protected]




























Bellovin                                                        [Page 4]


You are viewing proxied material from gopher.rbfh.de. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.