The MPLS WG Archive

[neil.2.harrison@bt.com]

Arun,

I think what Giles is trying to explain to us is like this (please correct
me if I am wrong here Giles):
-	where layer network trails have been standardised with operator
input there is OAM associated with the trails, eg ATM, SDH;
-	where they have not been standardised by the operators they don't,
eg ethernet;
-	so Giles is happy to make use of the OAM where its provided on an
end-end basis....
-	.....and where it's not he adds probe traffic, eg on
ethernet....which is a proprietary CV/P flow at the ethernet layer by any
other name;
-	and since there are always now end-end deterministic flows at the
client layers (eg OAM cell on ATM, probes on ethernet) then he can detect
any failures between end-end points....including the
ATM/ethernet/other<=>MPLS interworking function, and of course every single
node/link in all the layer networks between these points to the duct;
-	AFAIK he runs no OAM in the MPLS network since no one has provided
any yet...so can't do defect detection/diagnostics for problems that arise
there....but it also seems it he does not want it either;
-	he has proprietary metrics for availability/QoS monitoring based on
the higher layer provided OAM or the proprietary probe flows.

Does that help Arun?

I think the rest of us can now move on since Giles seems happy not to want
MPLS fault detection/handling (fair enough) and is clearly happy with his
own proprietary system/metrics which is running in the higher client layers
and has nothing whatsoever to do with MPLS.

regards, Neil

> -----Original Message-----
> From: Giles Heron [mailto:giles@packetexchange.net]
> Sent: 14 December 2001 15:19
> To: Punj, Arun
> Cc: neil.2.harrison@bt.com; mpls@UU.NET
> Subject: Re: MPLSOAM BOF meeting draft minutes
> 
> 
> "Punj, Arun" wrote:
> > 
> > Giles,
> > 
> > > -----Original Message-----
> > > From: Giles Heron [mailto:giles@packetexchange.net]
> > > Sent: Friday, December 14, 2001 5:18 AM
> > > To: neil.2.harrison@bt.com
> > > Cc: mpls@UU.NET
> > > Subject: Re: MPLSOAM BOF meeting draft minutes
> > >
> > >
> > > neil.2.harrison@bt.com wrote:
> > > >
> > > > Giles...you are so mysterious, some flesh on the bones of
> > > your hints might
> > > > help us all.  See below, regards Neil
> > > >
> > > > > I do not use customers as defect detectors.  As I stated
> > > > > in another
> > > > > post I have dedicated management servers for this.
> > > > NH=> So what do they do then?
> > >
> > > that is largely proprietary.
> > 
> > Would you say that something which you consider important 
> enough to go out
> > and implement
> > as a proprietary solution is not worth standardizing ?
> > 
> > In anycase, your logic is I already have it, so no one else 
> needs it.
> > Strange logic that.
> 
> that's not my logic at all.
> 
> We seem to have drifted off the original subject somewhat in any case,
> but if you read my other posts I think you might even start to
> understand that MPLS OAM will never be sufficient to manage PWs (since
> it injects CV packets inside the PW endpoints) and so something
> "proprietary" is required if you need to be sure that your PWs are up
> and if your PW payload protocol provides no such indication itself.
> 
> Giles
> 
> > Arun
> > 
> > <--snipped-->
> 
> -- 
> =================================================================
> Giles Heron    Principal Network Architect    PacketExchange Ltd.
> ph: +44 7880 506185              "if you build it they will yawn"
> =================================================================
>