The MPLS-OPS Archive

Cell Relay Retreat>MPLS-OPS Archive>month:2004-Aug> msg00166



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

Signalling a LSP

  • From: Alok <alokdube@hotpop.com>
  • Date: Thu, 26 Aug 2004 16:52:04 +0530
  • Resent-Date: Thu, 26 Aug 2004 08:16:04 -0400
  • To: mpls-ops@mplsrc.com
  • User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.2.1) Gecko/20030225
  • X-HotPOP: ----------------------------------------------- Sent By HotPOP.com FREE Email Get your FREE POP email at www.HotPOP.com -----------------------------------------------

Hello Folks,

can people please share some information on the best way to setup 
signalled LSPs?

Typically, are they as follows:
a. setup an LSP to an end point.
b. implement some CAR/rate-limited way to cap traffic to get to the end 
point.

at this point, as the signalling travels to the end point of the LSP, it 
has to satisfy some constraints.

However, say we have;

A-----B------<big network>--C--<big network>----D---E
                                                   

and i wish to setup an LSP which satisfies some constraints like 
bandwidth/Latency etc.

C being the common point for all possible paths from A to E.

Is there a way that:

a. The signalling carries the source of the LSP and say a "sequence" number.
b. The signalling tries all possible paths to E via the big network domain
c. the (sequence number, source identifier) are sufficient to identify 
the "signal" to avoid looping.
d. C may not be a looping point but since it has seen /tried to signal 
the (sequence number, source identifier) it does not try the same again.
e. again from C multiple paths are signalled (all of those which can 
satisfy the constraints)
f. one of them reaches E 1st, if there is enough (constraint) from D to 
E, D forwards it to E , else D can simply "drop/send a nack".
g. once E has accepted one path to A and sets up a reverse path, it can 
drop all other incoming based on source identifer/send a nack to all 
other requests that hit it based on (source,sequence).

-thanks
Alok





-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml