Last Call Review of draft-ietf-manet-dlep-25
review-ietf-manet-dlep-25-secdir-lc-hoffman-2016-11-17-00

Request Review of draft-ietf-manet-dlep
Requested rev. no specific revision (document currently at 29)
Type Last Call Review
Team Security Area Directorate (secdir)
Deadline 2016-11-28
Requested 2016-11-10
Authors Stan Ratliff, Shawn Jury, Darryl Satterwhite, Rick Taylor, Bo Berry
Draft last updated 2016-11-17
Completed reviews Genart Last Call review of -25 by Matthew Miller (diff)
Secdir Last Call review of -25 by Paul Hoffman (diff)
Opsdir Last Call review of -25 by Linda Dunbar (diff)
Rtgdir Early review of -13 by Lou Berger (diff)
Tsvart Telechat review of -25 by Michael Scharf (diff)
Assignment Reviewer Paul Hoffman
State Completed
Review review-ietf-manet-dlep-25-secdir-lc-hoffman-2016-11-17
Reviewed rev. 25 (document currently at 29)
Review result Has Issues
Review completed: 2016-11-17

Review
review-ietf-manet-dlep-25-secdir-lc-hoffman-2016-11-17

Greetings. This is a review of draft-ietf-manet-dlep-15 for the Security 


Directorate. Please treat these comments as you would any IETF Last Call 


comments you get.






As I understand it, Dynamic Link Exchange Protocol (DLEP) is a protocol 


for a router and wireless modem to inform each other about 


characteristics of the link in order to make better routing decisions. 


It runs over UDP and TCP, and is explicitly meant to be only valid on a 


single L2 hop directly between the modem and router. (Please let me know 


if I have this wrong!)




There is no security in DLEP. At the end of Section 3, it says:
   DLEP further requires that security of the implementations (e.g.,
   authentication of stations, encryption of traffic, or both) is dealt
   with by utilizing Layer 2 security techniques.  This reliance on


   Layer 2 mechanisms secures all DLEP Messages - both the UDP 


discovery



   Signals and the TCP control Messages.


Further, there is no mandatory-to-implement L2 security protocol; 802.1X 


and 802.1AE are mentioned as possibly being used, but neither is 


required to be implemented.






This, the specified security is pretty weak. It is not clear what 


advantage an attacker would get by snooping on the DLEP traffic: the 


values exchanged are pretty easy to determine just by watching the link. 


It is also not clear what advantage an attacker would get by 


impersonating either party or mounting an MITM attack other than 


degrading the link, which an MITM could do anyways.






This feels like a classic IETF "we don't deal with security and leave it 


to the layer below us" protocol, which might be acceptable in this case 


because of the nature of the data being exchanged.




--Paul Hoffman