Re: [Softwires] BGP TE attr last call by softwires WG (2nd question)

[Igor Bryskin <[e-mail]>]

Hi Adrian,
Please, see in-line.

Igor


----- Original Message ----
From: Adrian Farrel <[e-mail]>
To: Igor Bryskin <[e-mail]>
Cc: [e-mail]; [e-mail]
Sent: Monday, September 8, 2008 6:23:52 AM
Subject: Re: [Softwires] BGP TE attr last call by softwires WG (2nd question)

Hi Igor,

Not sure that the scope of this discussion is particularly appropriate to 
CCAMP and Softwires mailing lists. Might be more relevant on the L1VPN and 
OSPF lists. You could open a thread there if you want to debate this 
further.

IB>>As you remember it was not even me who started this discussion. I simply reacted to the following Yakov's statement addressed to the WGs :

> >And while on the subject of scaling, please keep in mind that BGP
> >only stores L1VPN routes on PEs that have sites of that VPN connected
> >to them, and on an RR if used, but *not* on any of the P routers. In
> >contrast, rfc5252 (OSPF for L1VPN autodiscovery) results in storing
> >*all VPN TE information for all the VPNs* on *all* the IGP nodes, both P
> >and PE. So, clearly BGP-based approach scales better than OSPF-based
> >approach.

In your reference below you seem to agree that this statement is not exactly correct. Now let me explain why I think that this discussion could be interesting to the CCAMP WG (and I will certainly annoy much more people now :=). *IF* GMPLS Control Plane (which includes OSPF and does not include BGP) can provide a viable alternative to L1VPN-BGP, then how about L2VPNs? This is important especially in the context of the MPLS-TP work, where GMPLS is the control plane of choice today, and it does not include BGP, but MPLS-TP still needs the L2VPN solution. So, should we extend GMPLS with BGP or this is not necessary?
Wouldn't this be a good discussion for CCAMP WG list? How about L2VPN, PWE3, MPLS TP lists? 


> Although multi-instance OSPF was enthusiastically discussed and is ongoing 
> work in OSPF WG, it was never discussed in the L1VPN context.

Please see RFC 5253 section 7.1

     Alternatively, a separate instance of
     the OSPF protocol can be used just between PEs for distributing
     membership information.

This appears to be exactly the point you are making. So perhaps we are all 
in agreement?

IB>> One thing what is missing IMO in this RFC (and is actually part of this discussion) is comparison of configuration efforts for BGP and OSPF solutions.
A solution may have good scalability properties but require enourmous configuration efforts, and thus become very expensive. Would you agree?

Thanks,
Igor

Adrian