[Softwires] Re: draft-ietf-softwire-hs-framework-l2tpv2-00 comments
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Softwires] Re: draft-ietf-softwire-hs-framework-l2tpv2-00 comments






5.1.2.  IPv4

   A Softwire Concentrator MAY provide either globally routable or
   private IPv4 addresses.  If private addresses are used, the delegated
   prefix should be in the same address space than the PPP endpoint to
   avoid nested NAT configurations.  A globally routable address is
   preferable, since in most cases, it is expected the CPE device will
   perform the IPv4 NAT function.

   The PPP link for the IPv4 softwire SHOULD be assigned a /30.

WFS Why a /30?

I imagine that since  it is a point to point link /30 allows to number both ends.
Do you think shorter prefixes should be allocated if user wants to have a network
instead of a host connected to the v4 Internet ? 

   Framing Capabilities AVP

      Synchronous bit MUST be set to 1 and Asynchronous bit to 0.  This
      AVP SHOULD be ignored by the receiver.

WFS I still think we should recommend to set both bits here.  They are
really meaningless in this context but being less restrictive should
increase our chance of interoperating.
Perhaps we should run this by Mark Townsley.

we can swap SHOULD and MUST

      Synchronous bit SHOULD be set to 1 and Asynchronous bit to 0.  This
      AVP MUST be ignored by the receiver.
 


   Challenge and Challenge Response AVPs

      Session authentication as defined in [RFC2661] is based on a
      shared secret known by LACs and LNSs, and is not designed to
      authenticate a specific user.  This AVP is not required since
      security enhancement is not guaranteed.  It can be used to limit
      DoS attack but since this secret has to be known by all users
      accessing the service, an attacker can learn it easily.

WFS I'm OK with not requiring tunnel authentication.  However, we should
remove
    the statement that the shared secret has to be known by all users.
    Both LAC and LNS much have the same secret, however that secret can
    be different for each LAC as long as each LAC has a different
hostname.

OK, we can  remove last sentence

      Session authentication as defined in [RFC2661] is based on a
      shared secret known by LACs and LNSs, and is not designed to
      authenticate a specific user.  This AVP is not required since
      security enhancement is not guaranteed. 
 

      While user authentication is typically done at the PPP level,
      tunnel authentication may be helpful in preventing DoS attacks.

6.2.2.  LCP

   Once the L2TP session is established, the SI initiates the PPP
   connection by sending a LCP Configuration Request message.  The SC
   also sends a LCP Configuration Request containing at least the
   Maximum Receive Unit option and and authentication protocol.  If no
   authentication protocol option is present, the SI considers the
   service as un authenticated (see Section 6.2.3).  Each party answers
   with a Configuration Ack message to finish the link configuration.

   ### Laurent, do you have an example for this section?

I will send you some dump of our traffic to specify options.


--
-----------------------
Serveur web du Point6 => http://www.point6.net
Laurent Toutain
ENST Bretagne
tel +33 2 99 12 70 26 - Fax +33 2 99 12 70 30
_______________________________________________
Softwires mailing list
Softwires at ietf.org
https://www1.ietf.org/mailman/listinfo/softwires

Note: Messages sent to this list are the opinions of the senders and do not imply endorsement by the IETF.