PCE Working Group Meeting - Thursday, November 11, 2010,
5:40 PM
- Co-chairs continue to work on
recharter text. Text will be provided for the WG to review.
- Two new RFCs since Maastricht (RFC 6006: PCEP Extensions for P2MP TE-LSPs and
RFC 6007: Synchronized VECtor List)
- Co-chairs are requesting a PCE Vendor Constraints contributor to help
progress the document.
- WG has 13 new documents, yet almost no discussion on the list. We need more private
discussions to be moved to the WG list.
- Co-chairs request contributors of new drafts to consider PCE architecture
implications when proposing new solutions.
2.1. PCEP
Extensions for GMPLS
Cyril Margaria
was the presenter.
- Authors were reminded that we should avoid proprietary
architecture for carrying vendor constraints. Also any experimental code point
requests should be limited to one or two code points.
- Authors clarified that load balancing is described in the document to handle different
route requests, not route diversity.
- Co-chairs highlighted that load balancing has no relevance to VCAT, so the
document terminology usage may need updating.
2.2. PCEP
Requirement for RWA
Presenter was Young Lee.
- Authors requested working group last call. Co-chairs pointed out
there are minor gaps in the document that need editorial updates and fixes. Authors
will perform review and editorial clean up and add text where necessary.
- Co-chairs requested that the authors of this document consider the general
requirements and WSON specific requirements and if necessary, sync requirements
with the GMPLS document authors.
2.3. PCEP
Extensions for Signal Compatibility Constraints in WSON
Presenter was Young Lee.
- WG highlighted that there is already a requirements document for
WSON. Authors mentioned the basic RWA requirements and OEO/Compatibility
requirements were previously discussed and it was decided to keep them
separate.
- WG also highlighted that OSPF typically carries the information described in
the document. Why would PCEP need to duplicate this information? Authors
mentioned that the IGP/OSPF conveys some information (possibilities) but
additional information is also required for the PCE, especially if multiple
choices exist. Offline discussion is suggested to clarify situation.
2.4.
Requirement for Cooperation between PCE and Distributed Routing Controller
- Solution described seems to be a distributed routing engine that
computes routes. Authors were reminded that the IETF specifies protocols
between network elements.
3.1.
Extensions to PCEP for Hierarchical PCEs
- Co-chairs highlighted that objective functions may benefit from
some description of the use case. This can be described in the H-PCE framework
document.
- It was established that 15-20 people in the room had read the H-PCE
extensions document.
3.2.
Shortest Constrained P2MP Inter-Domain TE LSPs
- Co-chairs requested the authors confirm the optimality of the
core-tree. Is the core-tree truly optimal, or simply provides a “good enough”
solution.
- WG discussions included how recovery should be handled. Perhaps
in the event of failure, computation of the path should be limited where
possible, so the entire should not be recomputed. Additionally the co-chairs
reminded the WG that local protection can be provided with fast reroute
(RSVP-TE) mechanisms and P2MP backup trees (NP hard).
- Co-chairs requested that the authors highlight protection and
recovery discussions and move them to the list.
Co-chairs propose that WG waits for one week. If there are no
objections then co-chairs will poll for WG adoption.
3.4. Relayed
CSPF for Multi-Area Multi-AS PCE
Cancelled
4.1.
Temporary Reservation of Computed Path Resources
- Co-chairs highlighted
their concerns with stateful conditions that may
cause race conditions in some scenarios. If resource priority is required, then
existing techniques already provide priority and soft pre-emption. It is
important that the use of PCE does not prevent other PCC requests to access
network resources.
- The authors presented a solution that will require further
discussion and clarification that the mechanisms described are necessary.
Further mailing list discussion is requested.
- Co-chairs requested that the authors consider and document
motivation of the solution in more detail. Careful consideration of synchronization
with multiple PCEs will also be required.
- No specific comments.
- Motivation of the mechanisms described requires further
discussion. It is not clear why existing solutions using ERO and XRO techniques
cannot be used for similar functionality.
- It was highlighted by a WG member that service backup is the responsibility
of the network planning department.