I agree that we do need an applicability draft for stateful PCE. Here are some initial comments on draft-zhang-pce-stateful-pce-app-04.
This draft is a good starting point for discussing the applicability of stateful PCE and covers a good set of use cases. Should this draft get adopted by the WG at some later stage, I think the definitions of "active stateful", "passive stateful" should be moved to here rather than in draft-ietf-pce-stateful-pce, along with the use cases. The draft abstract should also mention that this draft discusses the limitations of stateful PCEs. Section 2: The definition of minimum cut-set is wrong, because it states that the smallest set of links is the cut set. Change "the minimum set of links" to "the set of links with minimum capacity" and the definition is correct. Section 4.1 on coexistence. I wonder if you think that more than one active, stateful PCE can coexist in the same network? Would you need to partition the controlled PCCs between them? How would you keep their LSP databases synchronized with each other? Section 4.2. This section might be the best place for a discussion of the limitations of the stateful PCE's TE and LSP databases. Each of them lags behind the true state of the network, the amount of lag being determined by the channels by which the network updates the TE and LSP databases. Thus use of a stateful PCE reduces but does not eliminate the possibility of crankbacks, for example, and does not allow you to reach 100% utilization in a dynamic network (although it undoubtedly would improve network utilization in most scenarios). I think it is important to explain these limitations and the reasons for them, otherwise people might get the wrong impression that the stateful PCE is a perfect tool. As well as section 4.3 on survivability, which deals with recovery of the stateful PCE, you could also have a section 4.4 on redundancy. That is, multiple stateful PCEs in the network which all provide the same function to all clients should any one of them fail. Section 5.5 - I did not follow why the stateful PCE's interaction with the VNTM is any different that the stateless PCE. Are section 5.6 and 5.7 significantly different use cases? 5.7 read as a special case of 5.6. Suggest merging them. Typo spotted in section 5.7 "cam be used" -> "can be used". Regards Jon From: pce-boun...@ietf.org [mailto:pce-boun...@ietf.org] On Behalf Of JP Vasseur (jvasseur) Sent: 22 June 2013 09:20 To: Ina Minei; pce@ietf.org Subject: Re: [Pce] Stateful PCE applicability Thanks Ina - good question : WG, please voice your opinion Thanks JP. On Jun 21, 2013, at 9:16 AM, Ina Minei wrote: Dear chairs and working group, In light of the recent working group re-charter which now includes stateful PCE, we wanted to hear the opinions of the group on 1. the need for an applicability document for stateful PCE and 2. whether draft-zhang-pce-stateful-pce-app satisfies this need, or any gaps it might have Thank you, Ina and Xian
_______________________________________________ Pce mailing list Pce@ietf.org https://www.ietf.org/mailman/listinfo/pce
