Re: [mpls-tp] [PWE3] Proposal of using GAL for PW

[Alexander Vainshtein <[e-mail]>]

Dear Feng,
Lots of thanks for a comment. I fully agree with you that it is safe to use GAL for OAM in MPLS-PT environments because disallow usage of any form of ECMP is explicitly disallowed there.

What I have been trying to explain is that in more generic MPLS environments GAL-based OAM is not necessarily fate-sharing with the user traffic., while other OAM techniques are not affected by ECMP.

To me this is one more reason not to use GAL for PW OAM.

Regards,
     Sasha

From: HUANG Feng F [mailto:[e-mail]]
Sent: Tuesday, July 06, 2010 11:01 AM
To: Alexander Vainshtein; BUSI ITALO
Cc: [e-mail]; [e-mail]; [e-mail]; [e-mail]
Subject: RE: [PWE3] [mpls-tp] Proposal of using GAL for PW

Dear Sasha,
   In MPLS-TP transport network environment, because it is strict connected oriented, ECMP is not supported.
B.R.
Feng


________________________________
From: [e-mail] [mailto:[e-mail]] On Behalf Of Alexander Vainshtein
Sent: 2010年7月5日 20:42
To: BUSI ITALO
Cc: [e-mail]; [e-mail]; [e-mail]; [e-mail]
Subject: Re: [PWE3] [mpls-tp] Proposal of using GAL for PW
Italo,
Lots of thanks for a prompt response. However, IMHO it is not correct.

E.g., BFD for MPLS LSP (as defined RFC 5884) and BFD in VCCV (as defined in RFC 5885) will provide correct connectivity failure indications regardless of ECMP based on MPLS stack hashing because the same label stack will be used by user traffic and OAM packets.

Do you think I’ve missed something?

Regards,
     Sasha

From: BUSI, ITALO (ITALO) [mailto:[e-mail]]
Sent: Monday, July 05, 2010 3:29 PM
To: Alexander Vainshtein; [e-mail]; Shahram Davari
Cc: [e-mail]; [e-mail]; [e-mail]
Subject: RE: [PWE3] [mpls-tp] Proposal of using GAL for PW

Sasha,

what you say is correct but it is applicable to any environment where you use OAM together with ECMP. I do not see anything specific with GAL and PW.

Italo

________________________________
From: Alexander Vainshtein [mailto:[e-mail]]
Sent: lunedì 5 luglio 2010 10.39
To: BUSI, ITALO (ITALO); [e-mail]; Shahram Davari
Cc: [e-mail]; [e-mail]; [e-mail]
Subject: RE: [PWE3] [mpls-tp] Proposal of using GAL for PW
Hi all,
IMHO and FWIW, while RFC 5586  (5886 looks like a typo to me) does not place any restrictions on using GAL in non-TP environments, its usefulness in these environments (where various forms of multipath, including usage of entropy labels,  cannot be precluded) for OAM purposes is somewhat problematic:
Adding GAL to the label stack changes this for the multipath mechanisms based on stack hashing, and fate-sharing of OAM and user traffic packets is not guaranteed any more.

My 2c,
     Sasha

From: [e-mail] [mailto:[e-mail]] On Behalf Of BUSI, ITALO (ITALO)
Sent: Monday, July 05, 2010 11:17 AM
To: [e-mail]; Shahram Davari
Cc: [e-mail]; [e-mail]; [e-mail]
Subject: Re: [PWE3] [mpls-tp] Proposal of using GAL for PW

I agree with Stewart.

The GAL MUST be at the BoS in MPLS-TP where ECMP is not applicable and entroy labels MUST NOT be used.

In other MPLS environments (e.g., where entropy labels are used), RFC 5886 does not place any restriction on the position of the GAL within the label stack.

Italo

________________________________
From: [e-mail] [mailto:[e-mail]] On Behalf Of Stewart Bryant
Sent: venerdì 2 luglio 2010 11.24
To: Shahram Davari
Cc: Andy Malis; [e-mail]; [e-mail]; [e-mail]; HUANG, Feng F (Feng)
Subject: Re: [mpls-tp] [PWE3] Proposal of using GAL for PW
On 01/07/2010 19:35, Shahram Davari wrote:
!mmm so if one used entropy then they can’t have OAM?
-SD

Not so. The GAL goes above the entropy label (S=0), so gets parsed first. This works for both PHP and not PHP, and both LSP and PW.

- Stewart