draft-ietf-l3vpn-ospf-2547-04

[Eric Rosen <[e-mail]>]

As  a  result of  AD  review,  significant changes  have  been  made to  the
specification draft-ietf-l3vpn-ospf-2547.  These changes  can be seen in the
latest version, draft -04.  It is believed that the draft now corresponds to
the implementations.

The following issues were addressed as a result of the AD review.

The spec was  written so as to  allow a single VRF to  correspond to multiple
OSPF  domains.  However, it  did not  make clear  just which  parameters and
procedures are relative to a domain,  and which are relative to a VRF.  This
has  now been  cleared up.   However,  doing so  required extensive  textual
changes. 

There  are cases  where  BGP decides  to  put a  route into  the  VRF for  a
particular address prefix, and OSPF also decides to put a route into the VRF
for that same address prefix.  Of  course, only one of these can actually be
used for  forwarding.  The  original spec did  not make it  adequately clear
just how  a choice  between two such  routes would  be made.  This  has been
clarified.  In  some cases,  the results will  be different than  they would
have been if the VPN were really a pure OSPF network.  These differences are
now explained and their potential consequences pointed out. 

The  procedures  for  forwarding data  traffic  on  a  sham link  have  been
clarified.  The procedures  for sending OSPF control traffic  on a sham link
have been clarified.  The role of the optional  "sham link endpoint address"
has been clarified.

The  procedures for  translating BGP-distributed  VPN-IPv4 routes  into OSPF
routes have been clarified. 

A discussion  of NSSA routes has been  added.  Alex says it  is not detailed
enough; any feedback in this area would be welcome.

Due to the large number of changes,  Alex has asked for a new last call, and
I expect the WG chairs to formally issue the last call shortly.