Telechat Review of draft-irtf-rrg-recommendation-
review-irtf-rrg-recommendation-secdir-telechat-murphy-2010-10-29-00

I have reviewed this document as part of the security directorate's 


ongoing effort to review all IETF documents being processed by the IESG. 


These comments were written primarily for the benefit of the security area 


directors.  Document editors and WG chairs should treat these comments 


just like any other last call comments.






I unfortunately was off-net for a few days and got to this assignment 


rather late.  The document is long and covers a broad swath of material 


and I was not able to cover it deeply.






This document is a product of the rrg IRTF working group.  It summarizes 


15 different proposals for a new routing and addressing architecture for 


the Internet, with short summaries, critiques and rebuttals for each, and 


gives a final recommendation to the IETF for future direction.






With the breadth of scope of the document, there is no way for me to 


review each proposal's documents for security considerations.




The security considerations of *this* document itself is quite terse:

20. Security Considerations

   All solutions are required to provide security that is at least as
   strong as the existing Internet routing and addressing architecture.



Given the widely reported weakness of the "existing Internet routing and 


addressing architecture", this is a low bar indeed.  There are attempts in 


progress to attempt to improve the security of the Internet routing and 


addressing architecture.  I do not know what to suggest if these 


improvements leave the Internet with stronger security than is provided by 


these proposals.






The summaries of the different proposals devote little attention to the 


infrastructure security ramifications of the proposal.  Given the stated 


goal, perhaps no attention was necessary.






Many of these proposals include an encapsulation system, presenting the 


expected difficulties with end system authentication, filtering systems at 


boundaries, etc.  Some proposals addressed these concerns.  I am not sure 


if the security considerations section meant that the proposals were 


required to avoid weakening the end-host security protections already 


provided (ipsec, NAT, whatever).






The rrg wg came to consensus that a fundamental architectural feature is 


a separation of locator and identifier for any node.  Many of the 


discussed alternatives include a mapping system that produce a locator for 


a given destination identifier.






The mapping system would seem to be a very likely point of vulnerability, 


permitting traffic redirection for data exposure or blackholing, etc. 


Many proposals suggest a hierarchic architecture of the mapping system for 


scaling purposes.  I would presume that an authorization scheme for the 


mapping system would be essential, and that the hierarchy would be an 


important aspect of that scheme.  Of course, I can't tell much at this 


level of detail about how and if each proposals addresses this.  (One of 


the recommendations suggests communicating mapping info through bgp - I 


can not say at this point whether the SIDR suggestions for improving bgp 


security would be applicable.)




--Sandy

Nits:

   PMTUD  Path Maximum Transmission Unit Discovery: The process or
      mechanism that determines the largest packet that can be sent
      between a given source and destination with being either i)
      fragmented (IPv4 only), or ii) discarded (if not fragmentable)
      because it is too large to be sent down one link in the path from
      the source to the destination.



It should say "*without* being either", right?  A long sentence so I may 


have lost my place.







Several of the comments start using terms that are part of the wg 


deliberations, I'm sure.  But it makes reading the discussions and 


critiques obtuse.  In particular, "Core-Edge Separation" and "Core-Edge 


Elimination" seems to a well understood concept in the wg.  It needs to be 


defined somewhere.  A web search found references in some conference 


papers and in rrg mailing lists.