PCE Working Group Meeting

Meeting: IETF 91, Monday, November 10, 2014

Location: Hilton Hawaiian Village, Honolulu, Coral V, 1030-1130

Proxy Chairs: Dan King <daniel@olddog.co.uk>, Ramon Casellas <ramon.casellas@cttc.es>

Chairs in absentia: Julien Meuric <julien.meuric@orange.com>, JP Vasseur <jvasseur@cisco.com>

Minutes: Jonathan Hardwick <jonathan.hardwick@metaswitch.com>

1. Introduction

1.1. Administrivia, Agenda Bashing (chairs, 5 min)

1.2. WG Status (chair proxies, 15 min) [20/60]

draft-ietf-pce-enhanced-errors was last updated in Paris.  No authors present.  The group should check that this work is still alive.

draft-ietf-pce-pcep-stateful-pce-gmpls made more sense when the base stateful draft did not cover any GMPLS requirements, maybe now it is time to revisit whether this work is still required.

2. Already Discussed IDs

2.1. IRO and Domain Sequence (Dhruv Dhody, 5 min) [25/60]




Dhruv raised three questions.

1. Should the IRO survey draft be dropped or progressed to an informational RFC?

2. Is it possible to fast-track adoption and publication of the iro-update draft?

3. The domain-sequence draft is much simplified, are there any further comments from the working group on it?

No comments or questions from the floor.

2.2. DNS-Based PCE Discovery (Jeff Tantsura, 5 min) [30/60]


Jeff asked whether the WG could reconsider adopting this work?

Ramon: who has read the document? about 8 people raised their hands

Ramon: who thinks this is interesting to work on? a few less, maybe 6 people

2.3. Path Profiles (Rob Shakir, 5 min) [35/60]


Tariq from Cisco: are you considering making the profile ID a string rather than a number?

Rob: just planning it to be a number, would a string be useful?

Tariq: it will be useful to be able to give a name for the profile.

Tariq: how is the ID scope defined?

Rob: it is unique only between a PCC and PCE pair.  We are not so far making a case for globally well-known profiles

Dhruv: we have a different approach using Ina's associated LSP draft, which we think is more general, can we talk?

Rob: yes we will talk; our approach allows the operator not to have to configure a really complex path policy on the PCC.

Rob: please can we adopt this?

Dan: (after polling) plenty of people have read the document, a similar number of people think it is worth pursuing

3. New IDs not Discussed

3.1 PCEP BCP (Ramon Casellas, 5 min) [40/60]


Young Lee: This is good work, what will happen to the RFCs that violate this recommendation or else are unclear?

Cyril: We could make errata if there are any errors to fix.

Ramon: The RFCs will remain valid; we are not changing, only clarifying. This is an informational BCP.

Ramon: polls for who has read this (about 6), who wants to progress it (same)

Ramon: please use the mailing list if you want to see this proceed.

3.2. TE Data in PCEP (Young Lee, 10 min) [45/60]



Ramon: are you addressing TE data sync between PCEs?

Young: no, we have not discussed it yet, but it is in scope

Igor: what happens if you are getting information for the same link ID from PCEP, BGP-LS and/or IGP-TE?

Young: PCEP will usually get there first.

Igor: There should be a tie-breaker e.g. prefer one source over another, like you have in routing policy.

Wenhu Lu (Ericsson): You describe that it is up to the PCC what information to be send, I think it is better if this is configured on the PCE and requested via a pull mechanism, it is a more SDN-like model.

Ramon polls: who has read this (quite a lot), who thinks this is a good idea (about 10), who thinks this is NOT a good idea (again about 10)

3.3. Yang Model for PCEP (Dhruv Dhody, 5 min) [55/60]


Dan: it is important to understand the requirements, is YANG the right protocol for managing these devices?

Tariq: are you touching MPLS-TE specific items?

Dhruv: Not yet, TEDB and LSPDB will be generic things though.

Tariq: Agree, we should discuss how PCE will fit into wider picture of MPLS-TE.

3.4. PCEP Extension for Service Segment in SR (Qin Wu, 5 min) [60/60]


Qin: this is a small extension to the base draft-ietf-pce-segment-routing.

Rob Shakir: Too premature to discuss adopting this. No consensus that segment routing is the way to do service chaining.

??? (Ericsson): I support this draft as it is very simple and the existing SR draft says it supports the SR use case.

Ramon: We should move this discussion to the list.

3.5 Secure Transport of PCEP (Diego Lopez, 5 min)


No comments or questions from the floor.

Ramon closed the meeting at 11:30.