Minutes IETF114: pce: Thu 17:30
minutes-114-pce-202207281730-00
Meeting Minutes | Path Computation Element (pce) WG | |
---|---|---|
Date and time | 2022-07-28 21:30 | |
Title | Minutes IETF114: pce: Thu 17:30 | |
State | Active | |
Other versions | markdown | |
Last updated | 2022-08-02 |
IETF 114 PCE WG Minutes
- Chairs: Dhruv Dhody, Julien Meuric
- Secretary: Hari (Hariharan Ananthakrishnan)
- Slides
- Video
- Chat-Stream
- Meetecho Recording
PCE Working Group Meeting – Session I
Wednesday, July 27, 2022, 17:30-18:30 UTC
Introduction
1.1. Administrivia, Agenda Bashing (Chairs, 5 min) [5/60]
1.2. WG Status (Chairs, 10 min) [15/60]
1.3. State of WG I-Ds and next steps (Chairs, 15 min) [30/60]
- [Cheng Li] We do think SR-path-segment and SR-bidir is ready for
WG LC. - [Dhruv] Do we think we need to continue with both mechanisms in
the SR-path-segment draft? - [Cheng Li] Yes, unless there are comments in the mailing list.
- [Samuel Sidor] I wanted to update on the sid-algo doc. I got
feedback from multiple sources and there were some conflicting
opinions. I am trying to consolidate the updates which is causing
delay. We discussed between the authors and will do partial updates
and make progress faster. - [Dhruv] Wherever you are not able to make a conclusion, please add
an editors note in the document, so that WG can provide inputs.
Segment Routing
2.1 Circuit Style Policies (Samuel Sidor, 10 min) [40/60]
draft-sidor-pce-circuit-style-pcep-extensions-02
- [Adrian] If the path is delegated to PCE, but it is not allowed to
recompute, why was this delegated to PCE in the first place? - [Samuel] We still allow PCE to update the path but only if the
operator explicitly requesting it from the PCE. There is permanent
flag which means that PCE can still recompute but it will recompute
only if the original path is invalided. - [Adrian] You have a use-case that is driving this distinction for
you? - [Samuel] the main use-case is still keep the PCE delegation, just
to make sure that PCE doesn't change the path. - [Adrian] I hear you. I am not convinced
- [Dhruv] This is the second time this comment has come in. Its best
to add text in the document. - [Andrew Stone] PCE is allowed to tear down but not change the path
that way secondary can take over. - [Dhruv] We should be careful in what content we put in the SPRING
doc vs this. In spring doc where things that are PCE specific should
be moved to this document. SPRING document should not be specific to
PCE. You or Christian can discuss. - [Samuel] We can discuss any other separation would make sense.
- [Dhruv] I don't have strong opinion, but I want the WG to be
thinking about this. Should we think of as this as a policy or well
know path profile ? Should we do specific TLV/flags as proposed in
this doc. If this is the direction we are taking, we should be
cognizant. - [Andrew] We are trying to describe a policy in the doc, the flags
are not specific to circuit policy. These could be used for other
cases. - [Samuel] If anybody has different opinion we are discussing this
in the mailing list.
2.2 Redundancy Policy (Fan Yang, 10 min) [50/60]
draft-yang-pce-pcep-redundancy-policy-00
- [Dhruv] The SPRING WG redundancy protection I-D is SRv6 specific.
We have to make sure SR-MPLS for basic redundancy protection SID is
also supported. - [Fan] We can attach this for SR-MPLS and SRv6. I think we can
discuss and find a better way to fill the blank. - [Janos Farkas] Coordination with DetNet might be good.
- [Anthony] The Protection Type TLV diagram says 24 bits but the
diagram represents 32 bits. - [Dhruv] What is the difference between redundancy protection vs
protection we already have in PCEP ? How would it work with the
existing protection if someone specifies both ? - [Fan] We will come back with more clarification.
2.3 SID Verification (Tan Chen, 10 min) [60/60]
draft-chen-pce-sr-mpls-sid-verification-04
- [Samuel] I want to support the adoption. This draft is introducing
useful things. There are no other comments from the working group
for any major modifications. - [Dhruv] The new wiki page
https://wiki.ietf.org/group/pce/coordination is still WIP. We need
to do a better job in coordinating between the documents. Please
update the wiki, so that we can avoid issues with shared flags.
[Julien] See you tomorrow.
PCE Working Group Meeting – Session II
Thursday, July 28, 2022, 21:30-22:30 UTC
Stateful PCE
3.1 Operational Clarification (Mike Koldychev, 10 min) [10/60]
draft-koldychev-pce-operational-06
- [John Scudder] Mild preference for option 1 i.e. keep it in single
draft. - [Mike] Ok. Sometimes its hard to distinguish between information
vs standard. - [Andrew] Operational draft as checkpoint. Since there is some
update and clarification, I do like option 1. I do believe we should
adopt this. - [Dhruv] My preference would be two separate drafts. The separation
is hard to do. The draft states an update and dives into an example.
If its single doc, we should find a good way to explain. State the
normative first and then go on to explain. I will work with authors.
For an implementer this is easy, for external person this would be
harder to figure things out. - [Mike] Sure. In the beginning of the doc we say section x, y and z
are informational and others are normative. - [Julien] No one is recommending option - 3. We can ask for
consensus in WG between option 1 and 2.
3.2 IFIT (Giuseppe Fioccola, 10 min) [20/60]
- [Adrian] There was a draft in OPS area WG for IFIT framework. Your
draft is not making reference to the framework. I wondered if you
see any relationship or whether it is safe to go ahead without a
framework document guiding us. - [Giuseppe] We didn't refer because it was not adopted. May be the
framework needs to reference this one, since this can be seen as a
building block of the framework. - [Dhruv] The term IFIT needs to be clarified. By some mechanism we
need to solve this. There was one more comment during the adoption
call regarding the SR-MPLS in scope ? - [Giuseppe] We can omit SR-MPLS for now. Another thing to track.
- [Dhruv] Having MPLS in scope in important, if we can decide later
if we need to split into multiple documents later. - [Giuseppe] We have to follow the discussion there.
- [Dhruv] Since you have draft in IDR, from the start please take
care of coordination and keep the draft in sync.
DetNet
4.1 DetNet Bounded Latency (Quan Xiong, 10 min) [30/60]
draft-xiong-pce-detnet-bounded-latency-00
- [Janos Farkas] As Dhruv mentioned in the beginning, we are in
early stage of DetNet. We have lots of solutions in the table. As
for the information between PCE and devices, the flow info RFC can
be required. It might be good to wait until the work is more
matured. - [Andrew] How PCE learns the processing delay of routers in the
path? Is there another document ? - [Quan] The load-delay will be advertised before the path
computation. - [Andrew] Its good to clarify that in the document.
- [Lou Berger] In DetNet WG, we still have control plane framework
document is accepting contributions. If this work informs that or
any material that they like to add or see covered, it be great to
bring that in. - [Dhruv] My suggestion is to have clear requirements from DetNet
for PCEP and also clear consistent terminology between these various
solution drafts.
4.2 Enhanced DetNet (Li Zhang, 10 min) [40/60]
draft-zhang-pce-enhanced-detnet-00
- [Dhruv] All the comments given to the previous draft applies to
this one as well. In DetNet WG draft, there could be multiple BLIs,
but in PCE we have only one. Thus start with the clear requirements
from DetNet, that would be helpful in solution development.
Others
5.1 NRP ID (Quan Xiong, 10 min) [50/60]
- [Jie Dong] There is a similar draft called pcep-vtn, which
includes both the capability part and the LSPA extensions. There are
currently several control plane drafts (IGP, BGP, PCEP,...) on
network slicing extensions, the terminology used to be different.
Since there is some consensus about the network terminology in TEAS,
coordination among these documents would be needed, and we are open
to that. - [Quan] We proposed the draft very early. The difference between
the VPN plus and NRP:- may be we can discuss offline or in separate
meeting. - [Dhruv] Having a single document will be beneficial. I urge the
authors to work together.
#
Last updated 02 Aug 2022 at 0700 UTC