PCE Working Group Meeting – IETF 85

1. Introduction

1.1. Administrivia, Agenda Bashing (chairs, 5 min) [5/90]

1.2. WG Status (chairs, 10 min) [15/90]

- No Agenda changes

- Milestone update missed a number of goals.

- Document update (see WG chair slides)

- Author of PCE Vendor constraints highlighted that the document has a solution, and the authors are happy with it.

2. Stateful PCE

2.1. PCEP Extensions for Stateful Control (Ed Crabbe, 5 min) [20/90]

Co-chairs highlighted that mailing list discussion has been positive. They asked the working group if we should separate common procedures and technology specific extensions into multiple documents.
Author summarised that the stateful framework defines architecture and solution, with MPLS-TE application covered in a separate document. The authors continue to explore if and how the work should be split further for other technologies like GMPLS.
A comment at mic proposed a hierarchy model for PCEP functions and extensions.

2.2. PCEP Extensions for Stateful Control of MPLS-TE LSPs (Ed Crabbe, 5 min) [25/90]

2.3. PCEP Extensions for PCE-initiated LSP Setup in a Stateful PCE Model (Ina Minei, 10 min) [35/90]

A number of comments. These include that the PCC should always be able to revoke a delegation, look to reuse RBNF elements and inclusion of error codes and function.
Inter-domain was raised but the authors felt this is a topic for future discussion.
Authors indicated this is an early document and most of the points raised will be discussed and addressed in future versions.

2.4. PCEP Extensions for MPLS-TE LSP protection with stateful PCE (Ravi Torvi, 10 min) [45/90]

It was highlighted that the use of S-bit would overlap with GMPLs. Furthermore, it was proposed that the RSVP association object be used instead.
A suggestion that the proposal was applicable to stateless PCE as well as stateful PCE.
A scalability issue was raised that the proposal requires the PCE to have a session with each PLR in the network.
Authors confirmed that local control of facility backup is still possible.
A comment that the proposal changes existing PCE architecture by adding network control functionality. Authors responded that using PCEP is a cleaner solution to using RSVP-TE. Furthermore, the proposal is well within the scope of current PCE functionality (specifically path computation).

2.5. Open discussion (15 min) [60/90]

3. PCEP Extensions

3.1. PCEP for L3VPNs (Miyasaka-san, 10 min) [70/90]

This PCEP proposal requires the completion of additional RSVP-TE extensions.
Overall, interest in the concept but the solution may benefit from being applicable to existing RSVP-TE extensions and use cases.

3.2. PCE for Additional OF and Metrics (George Swallow / Zafar Ali, 10 min) [80/90]

A suggestion that P2MP should also be included. Also there is interest in bounded latency and lowest cost.

3.3. PCE for Explicit Inclusion or Exclusion of Abstract Nodes in P2MP (Dhruv Dhody, 10 min) [90/90]