I think it's okay because as the PW crosses the ECMP-enabled IP/MPLS domain
in the middle segment, you're no longer in an MPLS-TP environment and so the
GAL is not required to be BOS. During that middle segment, the PW flow
label would be placed below the GAL and above the GACh. It gets removed
when it hits the S-PE that switches you back into the MPLS-TP environment.
In other words, whether you're in an MPLS-TP environment is determined
segment by segment in a MS-PW.
Pablo
On Tue, Aug 16, 2011 at 1:02 PM, Alexander Vainshtein <
Alexander(_dot_)Vainshtein(_at_)ecitele(_dot_)com> wrote:
Hi all,
After having sent out my comments I've noticed that the specific example to
illustrate the need to combine GAL and "flow label" was inaccurate.
A more relevant example would look like following (I do not include a
diagram, but it can be easily provided if necessary)
1. A MS-PW:
- Starts at an S-PE that resides at the edge of an MPLS-TP domain
(no ECMP)
- Crosses this domain and enters an IP/MPLS domain with ECMP enabled
using a T-PE that resides at the age of these two domains
- Leaves this domain and enters a 2nd MPLS-TP domain (using the 2nd
T-PE)
- Terminates on another S-PE at the edge of the 2nd MPLS-TP domain
2. The operator intends to improve traffic distribution in the IP/MPLS
domain, hence he enables insertion and discard of "flow labels" at the
two S-PEs. Note that:
- This does not violate the MPLS-TP restriction on ECMP: ECMP does
not happen in he MPLS-TP domains
- T-PEs do not even have to be aware of flow labels
3. The operator also intends to operate some end-to-end OAM for this
MS-PW using "GAL-in-PW". This results in a conflict since both GAL and
"flow
label" are defined (in the corresponding drafts) as bottom of stack.
IMHO this describes a realistic scenario where the two drafts are in
controversy.
Regards,
Sasha
------------------------------
*From:* mpls-bounces(_at_)ietf(_dot_)org [mpls-bounces(_at_)ietf(_dot_)org]
On Behalf Of
Alexander Vainshtein [Alexander(_dot_)Vainshtein(_at_)ecitele(_dot_)com]
*Sent:* Tuesday, August 16, 2011 4:26 PM
*To:* ietf(_at_)ietf(_dot_)org
*Cc:* mpls(_at_)ietf(_dot_)org; Vladimir Kleiner; Idan Kaspit; Mishael
Wexler; pwe3;
Oren Gal; John Shirron; Rotem Cohen
*Subject:* [mpls] IETF Last Call comment on draft-ietf-pwe3-gal-in-pw
Hi all,
I would like to raise the following issue with regard to
draft-ietf-pwe3-gal-in-pw<http://datatracker.ietf.org/doc/draft-ietf-pwe3-mpls-tp-gal-in-pw/?include_text=1>:
controversy vs.
draft-ietf-pwe3-fat-pw<http://datatracker.ietf.org/doc/draft-ietf-pwe3-fat-pw/?include_text=1>with
regard to bottom-of-stack position.
As stated in the Introduction, this draft removes the restriction imposed
by RFC 5586 on usage of Generic Associated Channel Label (GAL) in PWs. The
corresponding text Section 4.2 of RFC 5586 states:
In MPLS-TP, the GAL MUST be used with packets on a G-ACh on LSPs,
Concatenated Segments of LSPs, and with Sections, and MUST NOT be used with
PWs. It MUST always be at the bottom of the label stack (i.e., S bit set
to 1).
draft-ietf-pwe3-gal-in-pw proposed to replace the original text in RFC 5586
with the following
In MPLS-TP, the GAL MUST be used with packets on a G-ACh on LSPs,
Concatenated Segments of LSPs, and with Sections, and MAY be used with PWs.
It MUST always be at the bottom of the label stack (i.e., S bit set to 1).
I.e., while removing this restriction of 5586, it does not modify its
requirement for the GAL being always at the bottom of the label stack.
At the same draft-ietf-pwe3-fat-pw (currently also in the IESG review)
reserves the bottom of the PW stack for the PW flow labels, e.g., in Section
1.1:
This document describes a method of adding an additional label stack entry
(LSE) at the bottom of stack in order to facilitate the load balancing of
the flows within a PW over the available ECMPs.
One could argue that draft-ietf-pwe3-gal-in-pw only applies to MPLS-TP
pseudowires, and that MPLS-TP does not use ECMP. IMHO and FWIW,
such an argument, were it presented, would be highly problematic, because:
1. RFC 5960 (which defines the MPLS-TP data plane) did not define
any differences between the PW data plane in IP/MPLS and MPLS-TP.
2. One of the most popular scenarios for using multi-segment
pseudowires is the case when an edge-to-edge service emulation crosses
multiple IP/MPLS and MPLS-TP domains. In these scenarios, the flow label of
draft-ietf-pwe3-fat-pw (inserted by a flow-aware T-PE at the edge of an
IP/MPLS domain) would potentially compete with GAL (inserted by a T-PE at
the edge of an MPLS-TP domain, e.g., for relying a PW status message that it
has received over a Targeted LDP session from the IP/MPLS domain to a static
PW status message to cross the MPLS-TP domain) for the bottom-of-stack
position.
The issue I am raising Is not new. It has been actively discussed on the
PWE3 mailing list with regard to adoption of draft-nadeau-pwe3-vccv-2 as a
WG document, with arguments for both the flow label and GAL taking the
bottom-of-the-stack position. But, to the best of my understanding,
consensus on this issue has not been reached.
Hopefully this comment will be useful.
Regards,
Sasha
This e-mail message is intended for the recipient only and contains
information which is CONFIDENTIAL and which may be proprietary to ECI
Telecom. If you have received this transmission in error, please inform us
by e-mail, phone or fax, and then delete the original and all copies
thereof.
This e-mail message is intended for the recipient only and contains
information which is CONFIDENTIAL and which may be proprietary to ECI
Telecom. If you have received this transmission in error, please inform us
by e-mail, phone or fax, and then delete the original and all copies
thereof.
_______________________________________________
pwe3 mailing list
pwe3(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/pwe3
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf