[07:13:18] --- dfedyk has become available
[07:13:26] --- dfedyk has left
[14:29:18] --- javier has become available
[14:31:09] --- Bill has become available
[14:33:42] --- dinakar has become available
[14:33:57] --- amalis has become available
[14:34:59] --- sadlerjon has become available
[14:35:18] <sadlerjon> JP is talking abiout what happened in San Diego
[14:35:19] --- dinakar has left
[14:35:35] <sadlerjon> Computing path over multi-area/domain
[14:35:51] <sadlerjon> Diverse path computation
[14:35:59] <sadlerjon> Complex constraints
[14:36:08] <sadlerjon> These are all difference problems that can be solved
[14:36:48] <sadlerjon> Complex constraints are NP complete problem, so off-line computation is necessary.
[14:37:27] <sadlerjon> Comon theme at SD PCE BOF was need for (G)MPLS TE LSP path computation
[14:37:39] <sadlerjon> Need to develop a draft charter
[14:38:12] <sadlerjon> Adrian points out that Keriti wants MPLS on JP's slides changed to GMPLS
[14:38:22] <sadlerjon> JP made a realtime change ;)
[14:39:34] <amalis> Adrian: CCAMP will commit to provide requirements to PCE
[14:40:11] <amalis> Adrian has reservations that the problem statement still needs work.
[14:40:43] <amalis> Jerry Ash presents draft-ash-pce-architecture-00.txt
[14:41:11] <amalis> Jerry presenting terminology from the draft
[14:41:23] <amalis> Reading verbatim
[14:43:08] <amalis> Reading his assumptions from his draft
[14:43:47] <amalis> (He did ask if he had 3 hours to do his presentation...guess he was serious...)
[14:43:48] --- sadlerjon has left: Disconnected
[14:44:02] <amalis> BTW - I'm covering for Jon while he's rebooting
[14:44:41] <amalis> Now he's reading his slides for the motivations for the PCE architecture
[14:45:48] <amalis> Jerry has a picture of a modular breakdown of a composite PCE node
[14:46:01] <amalis> Looks like ASCII art right from the draft
[14:46:51] <amalis> Jon's almost back in business ....
[14:47:14] <amalis> Jerry's now presenting PCE architectural considerations
[14:48:02] --- sadlerjon has become available
[14:48:04] <amalis> Some interesting issues are synchonization, PCE discovery and load balancing, PCE-PCE and PCC-PCE communications ...
[14:48:14] <amalis> And I hand it back to Jon.
[14:48:31] <sadlerjon> Back from my reboot (fine M$ product -- ugh)
[14:49:03] <sadlerjon> Plicy and confidentialliy issues must be handled in the information returned by the PCE
[14:49:26] <sadlerjon> ERO segment can be replaced with a cookie, keeping confidentiallity.
[14:49:57] <sadlerjon> PCC-PCE communications needs to handle "usual" security issues
[14:50:09] <sadlerjon> Snooping not a significant issue, can be handled through encryption
[14:50:20] <sadlerjon> Spoofing is a significant issue as it would allow for traffic intercept.
[14:50:59] <sadlerjon> PCE evaluation Metrics are as specified in the draft (optimality, scalability, load sharing, multiple path computation, reoptimization, path computation time, network stability, synchronization)
[14:51:20] <sadlerjon> JP says that Manageability is probably also something that should be added (wink at Adrian ;) )
[14:51:27] <sadlerjon> Issues raised on list
[14:51:51] <sadlerjon> PCE should advertise its capabilities (constrants that can be handled, etc.)
[14:52:12] <sadlerjon> Request should handle near-disjoint as well as manditary disjoint
[14:52:27] <sadlerjon> TED can include info from sources other than IGP
[14:52:41] <sadlerjon> Eval metrics should include TED sdync speed, and impact on data flows
[14:53:16] <sadlerjon> Arch should elaborate on advantages of stateful PCE and pitfalls of stateful PCE in a distributed PCE environment
[14:53:22] <sadlerjon> Questions at Mic:
[14:53:34] <sadlerjon> Mark: what is exactly expected to be standardized here
[14:54:05] <sadlerjon> Dimitri: Standardizing an entity, not a protocol. So what is expected to be standardize here? LSR behavior internally is not standardized.
[14:54:27] <sadlerjon> Adrian clarification: if PCE is colocated with LSR, what is the work to be done?
[14:54:44] <sadlerjon> Dimitri: Right
[14:54:52] <sadlerjon> <someone> <some comment>
[14:55:16] <sadlerjon> <someone else> does arch allow for PCEs to initiate rerouting in the network?
[14:55:30] <sadlerjon> JP: Stateful PCE may allow such behavior
[14:55:52] <sadlerjon> JP now onto charter discussion
[14:56:14] <sadlerjon> Sorry, onto discussion of existing and new drafts
[14:56:44] <sadlerjon> Mang out there, including interdomain framework, Procedural and Operational considerations for PCE in interdomain
[14:56:50] <sadlerjon> Use of PCE for MPLS Fast Reroute
[14:57:02] <sadlerjon> GMPLS considerations
[14:57:16] <sadlerjon> Draft proposing specific PCE request protocol
[14:57:35] <sadlerjon> Since last IETF, there have been 8 new drafts -- shows interest in the area.
[14:57:46] <sadlerjon> JP now onto Key Questions...
[14:58:06] <sadlerjon> Clear requirements have been expressed by many Service Providers during the last BOF in San Diefo, on the mailing list, etc.
[14:58:46] <sadlerjon> JP requests comments on this first point
[14:58:47] <sadlerjon> Non made
[14:58:57] <sadlerjon> Next point: Does this work belongs to the IETF? Its under the IETF scope of expertise...
[14:59:25] <sadlerjon> JP points out scope is MPLS-TE LSPs, and looks toward Alex.
[14:59:37] <sadlerjon> Alex asks if there are any comments on JPs question
[15:00:08] <sadlerjon> <someone> Can't answer question as we havn't described what is to be standardized
[15:00:40] <sadlerjon> JP responds that the charter slide does provide more definition, but has not been presented yet. So ordering of slides was probably suboptimal.
[15:01:06] <sadlerjon> Adrian says Lets go to charter and we'll go back to the issues.
[15:02:16] <sadlerjon> Charter is on the web
[15:02:18] <sadlerjon> Milestones:
[15:02:32] <sadlerjon> - Specification of PCE-based architure
[15:03:15] <sadlerjon> - In cooperation with protocol WGs (OSPF, ISIS, IDR, MPLS, CCAMP) development of routing and LSP signaling extensions required by PCE-based path computation techniques.
[15:03:45] <sadlerjon> Signalling extensions is for a reply/response protocol, not for other changes
[15:04:16] <sadlerjon> Currently 6 protocol extension approaches on the tables...
[15:04:25] <sadlerjon> Routing extensions are for auto discovery of PCE
[15:05:00] <sadlerjon> Loa: 2 comments
[15:05:31] <sadlerjon> Loa: Has problems parsing the Second Milestone presented -- does not include a statement of what the goal is.
[15:05:35] <sadlerjon> How is this a milestone?
[15:06:01] <sadlerjon> Loa states that an action verb is needed
[15:06:15] <sadlerjon> Adrian says it suffers being translated from French to English to Swedish
[15:06:57] <sadlerjon> Loa: Second comment -- what is the limit to the extensions? Will Routing extensions to PNNI be proposed? Wants specific exclusion of LDP extensions.
[15:07:23] <sadlerjon> Adrian asks if Loa wants a specific list of excluded protocols.
[15:07:29] <sadlerjon> Loa says inclusion list is more complete
[15:08:22] <sadlerjon> - Specification of technieques in support of PCE discovery within and across domains. Where such techiques result in extensions of existing protocols, this work will be in conjunction with protocol WGs.
[15:08:43] <sadlerjon> - Specification of new communication protocol for use between a PCC (Path Comp Client) and a PCE and between PCEs.
[15:09:10] <sadlerjon> - Definition of protocol-independend metrics defining path quality measurement criteria and scalability criteria
[15:09:18] <sadlerjon> Specification of requirements for handling policy
[15:09:32] <sadlerjon> (Sorry, that was another Milestone)
[15:09:56] <sadlerjon> JP asks for comments:
[15:10:39] <sadlerjon> Arti: Asks if extensions for policy, security and confidentiality are specifically for diverse path computation.
[15:10:44] <sadlerjon> JP: No general requirement.
[15:11:55] <sadlerjon> Mark: Techniques are discussed alot in the drafts, why should we be specifying more than protocols?
[15:12:15] <sadlerjon> JP: Technique is not specified, but inputs/outputs need to be useful
[15:13:13] <sadlerjon> George S: Shouldn't specify algoritm being used/using PCE function
[15:14:06] <sadlerjon> Choi: Should be able to handle devices that don't want to include Signalling or ROuting capability. Adrian clarifies that PCE allows for such a thing
[15:15:23] <sadlerjon> Kiretti: Use of distirbuted or centralized computation should be an implementation detail -- not a standardized item.
[15:16:31] <sadlerjon> Kiretti: Shouldn't standardize the algorithm that says how PCE function is to be used
[15:16:59] <sadlerjon> Adrian: In the case of Multi-AS, isn't that distributed computation, and shouldn't that be standardized?
[15:17:36] <sadlerjon> Kiretti: The behavior of a PCE should be standardized, but the way that the PCE is used to develop an end-to-end route is not necessarily something that should be standardized.
[15:18:47] <sadlerjon> Anna C: Do we need to standardize the inter-PCE communication method?
[15:18:52] <sadlerjon> Adrian: Somewhat
[15:19:51] <sadlerjon> Richard: Distributed algoritm (not hierarchical decomposition) is an implentation decision, so it shouldn't be standardized.
[15:20:06] <sadlerjon> JP -- More Milestones discussion:
[15:20:33] <sadlerjon> Also includes:
[15:20:36] <sadlerjon> - Definition of MIBs and management procedures
[15:21:07] <sadlerjon> JP asks should CCAMP provide requirements?
[15:21:22] <sadlerjon> Kiretti: Making us pay for our previous transgressions requiring requirements, eh
[15:21:47] <sadlerjon> Kiretti: CCAMP will have some requirements to include, but other WGs also need to be consulted (such as MPLS)
[15:22:13] <sadlerjon> JP asks George if this is what he thinks, including handling things like pt2mpt LSPs.
[15:22:27] <sadlerjon> George: still have work to do, so can't specify all requirements at this time
[15:22:42] <sadlerjon> Says that unicast is probably good first step
[15:23:24] <sadlerjon> Adrian clarification: These milestones are not to cause another w/g to do significant work.
[15:24:14] <sadlerjon> Dimitri: Previous MPLS work was on signalling/routing protocols, not behaviors. PCE work should still not be about behaviors.
[15:24:37] <sadlerjon> Kiretti: Make certain that mpt2mpt LSPs are out of scope (causes him to Shiver)
[15:24:58] <sadlerjon> JP now on to discussion of initial milestones:
[15:25:03] <sadlerjon> Submit requirements draft
[15:25:21] <sadlerjon> submit draft describing the PCE arch
[15:25:35] <sadlerjon> Submit draft specifying communications protocol requiremetns between PCC and PCe and PCEs
[15:25:53] <sadlerjon> Submit draft of a SINGLE communication protocol for use between a PCC and a PCE and between PCEs
[15:26:00] <sadlerjon> Submit an applicability draft
[15:26:11] <sadlerjon> Kiretti: Is a MIB required?
[15:26:27] <sadlerjon> Adrian: If we have a protocol, a MIB will be required. Not immediately obvious
[15:26:42] <sadlerjon> JP: We already have a volunteer for a MIB (Tom ;) )
[15:26:59] <sadlerjon> Kiretti: Agrees we need management, but questions whether a MIB is needed.
[15:27:12] <sadlerjon> Adrian: says it is required by IESG...
[15:27:47] <sadlerjon> Richard: Does this include Load ballencing between PCEs?
[15:28:19] <sadlerjon> JP: MPLS PCE Load ballencing draft has been created already, need to have GMPLS extensions
[15:28:43] <sadlerjon> Richard: Doesn't this mean that PCE needs to provide information on its Compute power, etc? Does that cause confidentiality problem?
[15:28:49] <sadlerjon> JP: Ways around this...
[15:29:19] <sadlerjon> Kiretti: Wants to walk before we run, but would like to see charter include support for L1VPNs
[15:29:38] <sadlerjon> JP back to key questions.
[15:29:47] <sadlerjon> Alex asks if there are any questions on the key questions.
[15:29:55] <sadlerjon> (Other key questions listed are:
[15:30:05] <sadlerjon> - Is there enough interst on this architecture?
[15:30:16] <sadlerjon> CCAMP consensus states yes
[15:30:23] <sadlerjon> Ready to create a new WG?
[15:30:24] <sadlerjon> )
[15:30:35] <sadlerjon> Alex: there are two questions:
[15:30:48] <sadlerjon> - Is there a need for a WG? Many hands raised
[15:31:06] <sadlerjon> Are there any people that don't believe there is a need for a WG? No hands raised
[15:31:31] <sadlerjon> Alex: Hmm Not all hands raised for those two questions. Who isn't thinking? Some hands
[15:31:47] <sadlerjon> Ok. Seems to be concensus in the room.
[15:31:52] <sadlerjon> Alex: will take it to the IESG for consideration.
[15:32:03] <sadlerjon> Keep discussion going on PCE mailing list.
[15:32:09] <sadlerjon> Thanks for coming
[15:32:14] <sadlerjon> JP: Thank you!
[15:32:33] <sadlerjon> </BOF_meeting>
[15:32:48] --- amalis has left
[15:32:54] --- sadlerjon has left
[15:32:56] --- javier has left
[15:58:37] --- paul.knight has become available
[16:01:05] --- paul.knight has left
[19:58:34] --- Bill has left