The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2001-Jul> msg00292



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

[Fwd: I-D ACTION:draft-pan-lsp-ping-00.txt]

  • From: Sanjay Wadhwa <swadhwa01@yahoo.com>
  • Date: Wed, 18 Jul 2001 07:54:02 -0700 (PDT)
  • Cc: "Punj, Arun" <Arun.Punj@marconi.com>, mpls-list <mpls@UU.NET>


--- Eric Rosen <erosen@cisco.com> wrote:
> 
> Ping> The problem with this approach is: there is no guarantee that
> the
> Ping> returning path (with or without IP Router Alert) won't go over
> an LSP.
> Ping> If that LSP is bad, we are in trouble.
> 
> Since tunnels can be nested, isn't is possible that in a traffic
> engineered
> path <R1,  R2, R3,  ..., Rn>,  R2 and R3,  say, are  IGP adjacencies 
> over a
> virtual interface that  is actually a pair  of LSPs (one from R2  to
> R3, one
> from R3 to R2)?  In such a  case there is no guarantee that the
> control path
> (forward and reverse) does not include an LSP. 
> 

Sure. But if the reverse LSP (virtual interface) is  broken (say due to
corrupted label maps along the way), then even the RESVs for the inner
LSPs don't make it through, and soon enough the inner LSP will be torn
down (So making echo-response live and die with rsvp resv actually will
work fine with nested tunnels irrespective of RESV using a reverse LSP
or native ip in the reverse direction).

-Sanjay

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail
http://personal.mail.yahoo.com/