1. Agenda bashing (none)
2. WG Status
Adrian F. is new Routing Area AD and AD responsible of MPLS WG
Martin V. Secretary of MPLS WG
New RFCs since IETF73rd
- RFC 5439 "An Analysis of Scaling Issues in MPLS-TE Core Networks"
- RFC 5443 "LDP IGP Synchronization"
- RFC 5462 "Multiprotocol Label Switching (MPLS) Label Stack
Entry: EXP Field Renamed to Traffic Class Field"
Working Group drafts:
a. draft-ietf-mpls-explicit-resource-control-bundle
should be ready for IESG
b. draft-ietf-mpls-fastreroute-mib
(on the agenda)
c. draft-ietf-mpls-ip-options
ready for WG LC
d. draft-ietf-mpls-mpls-and-gmpls-security-framework
moving forward, proto write-up being written by Martin
e. draft-ietf-mpls-p2mp-lsp-ping
needs to be updated. Should be available in a few weeks.
f. draft-ietf-mpls-remote-lsp-ping
George to update and then to LC
need Security AD review in parallel with WG LC
g. draft-ietf-mpls-rsvp-te-no-php-oob-mapping
needs more work before going to WG LC
h. draft-ietf-mpls-ldp-typed-wildcard (Expired)
very small update needed to be done
i. draft-ietf-mpls-3209-patherr
draft-ietf-mpls-gmpls-lsp-reroute
draft-ietf-mpls-soft-preemption
Publication requested
j. draft-ietf-mpls-p2mp-te-mib
document is in a second working group last call.
k. draft-ietf-mpls-ldp-capabilities
In IESG Evaluation
l. draft-ietf-mpls-ldp-end-of-lib
LC has ended
m. draft-ietf-mpls-lsp-ping-enhanced-dsmap
Nitin: document is stable
Loa: next revision ready for LC?
Nitin: yes
n. draft-ietf-mpls-ldp-p2mp (Expired)
o. draft-ietf-mpls-ldp-upstream (Expired)
p. draft-ietf-mpls-rsvp-upstream (Expired)
Loa: have expired. Will try to find another co-author.
IPR claims
Ross: authors aware of ipr should disclose that. There is a
process for that. If IPR exists, it is up to the wg to decide
what to do with a draft with IPR claims.
General status of MPLS-TP
Loa: We are doing progress. Some drafts are ahead of schedule
while others are behind, but overall we are having good
progress. IPR claim on MPLS-TP Framework regarding ring
protection (from ECI). ITU-T is asking for a schedule of stable
drafts and RFCs. IETF is asking for the list of documents to
prioritize.
Malcolm: I should have the list by this evening
3. draft-ietf-mpls-fastreroute-mib (Tom)
draft, slides are missing
Wrapping up after 4 years. LC a year ago. Several exchanges with
MIB doctors since then. Still addressing a few MIB doctor
comments. Several implementations of the MIB exist. They have
shared experience and feedback. Second LC will be requested after
next revision.
4. draft-so-yong-mpls-ctg-framework-requirement (Ning So)
draft, slides
Added several requirements following feedback and input of new
co-authors. Next step is to request for RTGWG adoption
George: a very similar document is in progress at the ITU.
So: yes, G.800
George: how does it relate to this doc?
Ning
: ITU work is finished. ITU-T and IETF could work together on
CTG.
Loa: is this work fully aligned with existing work of MPLS?
Ning
: For the moment, only requirements. The objective of the draft
is to state carriers, perspective on requirements and framework.
The way these requirements should be implemented is outside the
scope.
Loa: have you agreed with RTGWG that they would to the write-up?
Ning
: yes
Andy: this document will be taken up by rtgwg
Loa: ok, please make sure that it is also reviewed by the appropriate
WGs as the doc progresses.
Lou: relation with Link Bundling?
Ning
: The draft describes specifically the differences between the
two. CTG provides major advantages: can present the group as a
single virtual interface, can group different type of interfaces
and different speeds.
Lou: if different interface types how do you do the advertisement?
Ning
: described in the draft
Lou: we should then decide whether we deprecate Link Bundling
Loa: needs to be discussed
Andy: The draft is the result of operational concerns we had with
link bundling.
Lou: I am concerned to see issues and the answer is a new solution
as opposed to let's try to fix the existing one.
Loa: WG does not need to take any action.
5. draft-ali-mpls-inter-domain-p2mp-rsvp-te-lsp (Nic)
draft, slides
Nic presented.
Loa: who has read? Less than a hand-full. Take to the list.
6. draft-pdutta-mpls-tldp-hello-reduce (Tom)
draft, slides are missing
The intension of the draft is to standardize a mechanism to
optimize T-LDP scaling. After the T-LDP session is established,
Hellos are redundant.
Loa: need to understand the real problem. Is hello traffic really a
problem?
Tom: Yes. This just to removes the unnecessary hellos.
Loa: understood, but how big is the problem?
Tom: it is a problem on one of our deployment.
George: I think the draft is fine, but maybe values other than
infinity should be allowed.
Tom: Fine.
George: Why it is standard track and not informational?
Tom: KInformational is fine.
Luca: agree good thing to do, but why not get rid of Hellos
completely?
Tom: we thought about it but not decided, that would be a
significant change.
Nitin: isn't 10 seconds sufficient?
7. draft-lu-ldp-igp-sync-bcast (Sriganesh)
draft, slides
LDP IGP Sync is an RFC 5443. But 5443 does not address broadcast
networks. What ever policy is chosen (divert traffic or not divert
traffic) (see section 3) traffic may result in some traffic being
blackholed. Proposed solution: remove link from lsdb unless
absolutely necessary (i.e. when link is a cut-edge) until sync is
complete.
Greg M.: addition of new neighbour would affect stability on
existing nodes connected to broadcast. Nodes will see new
neighbour but no ldp session and you say link not up until ldp
connection up.
Sri: no, procedure only applies to nodes with no (pre-)existing
adjacencies.
Greg: but how long to stabilize the system
Sri: no issue with stabilization
Greg: lets have an infinite number of nodes on segment, what if ...
George: very interesting discussion. Please continue off-line.
8. draft-ietf-mpls-tp-requirements (Ben)
draft, slides
Should be ready to be sent to IESG after next revision (06). Could
send to ITU at same time?
Loa: Malcolm are you okay with this procedure?
Malcolm: expecting that 06 goes to WG LC, not straight to the ITU,
2 week email review, a couple of days for the liaison crafting
should be okay. Didn't see anything on the list, so expectation
is low for major changes.
9. draft-ietf-mpls-tp-framework (Matthew)
draft, slides
Proposal to maintain this document as a living document until
MPLS-TP matures
no question/discussion
10. draft-ietf-mpls-tp-oam-requirements (Martin)
draft, slides
Next revision should be ready for LC.
Loa: Malcolm, Dave, is doc mature enough?
Malcolm & Dave: yes, doc is in good shape.
11. draft-busi-mpls-tp-oam-framework (Italo)
draft, slides
Loa: need to converge on TCM definition
Nurit: we are working on some text, we'll provide it, note that
TCM is more than just about OAM.
Eve: TCM is very well known in ITU.
Loa: not talking about this, but about terminology
Italo: you do not like the name?
Loa: concept well understood. need to work on name
12. draft-mansfield-mpls-tp-nm-framework (Scott)
draft, slides
no question/discussion
13. draft-sprecher-mpls-tp-survive-fwk (Nurit)
draft, slides
Nic: what is the mechanism used for the protection?
Nurit: we don't provide the solution, only the requirements.
Rahul: Suggest to include a reference to RFC4090.
14. draft-sprecher-mpls-tp-oam-analysis (Nurit)
draft, slides
Loa: Draft has expired, but we want to discuss this. can we do it?
It was on the agenda for the last meeting, but we didn't get to
it.
room: yes
Loa & george: allowed.
Nitin: Diagnostic function is now a MAY in oam-reqs so this draft
should not require new tool.
Nurit: OK
Nitin: doc states that lsp ping can not be used. clarification on
why should be provided.
15. draft-weingarten-mpls-tp-linear-protection (Yaacov)
draft, slides
Adrian: linear protection does not appear in 4427 so referencing
it is awkward. Concern for a new protocol should seek
consistency between MPLS-TP and GMPLS
George: consistency between the GMPLS and this work is very
important.
Nurit: okay
Italo: no concern about GMPLS. for in-band, G8031 must be used.
(?): uni-directional lsp considered?
Yaacov: we protect unidir and bidir but 1+1 only for unidir
Rahul: why document does not only focus on LSP?
Ben: There are resiliency reqs for LSPs and PWs
Rahul: does the reqs document clearly spell out which reqs apply
to server layer and which apply to server & client?
Ben: yes
Greg: do the same mechanisms apply to bidir and co-routed?
Yacoov: yes, think so. Will take the comment.
Nurit: just clarification that linear can work at any level and on
segments
16. draft-andersson-mpls-tp-process (Loa)
draft (no slides used)
Loa presented a potential planning/schedule of draft progress.
There may be an issue w/ multiple LCs in parallel. Draft depicts
process in ietf. Encourages ITU-T to do same thing and write
similar document. Once both pieces are available we can agree on
a global process.
17. draft-ietf-mpls-tp-gach-gal (Stewart)
draft, slides
Loa: authors have done well to take into account LC comments.
Will be sent to ITU-T for a final ACK.
18. draft-ietf-mpls-tp-nm-req (Eric)
draft, slides
Accepted as wg doc. Some discussions around the document but very
few really related to nm. Target is to publish by end of Sept.
Loa: Goal fits with the schedule.
19. draft-abfb-mpls-tp-control-plane-framework (Luyuan)
draft, slides
Objective: provide context for the MPLS-TP control plane.
Reminder: MPLS-TP does not require a CP so the current role of the
document is to drive which functions need to be in oam and which
in the CP.
Igor: good document and starting point. MPLS-TP allows to do
dynamic and static. suggest to incorporate NM/CP connection
ownership.
Lou: please raise this on list to see if providers agree with that.
Igor: suggest to remove :"mpls and mpls-tp cp should interop". On
CP for PWs, add rfc references and document should not say whether
it is ldp or bgp.
Rahul: why the end-to-end PW on slide 5?
Luyuan: it is in the overall framework
Rahul: relates to Monday comment; what is the server layer, the
client layer?
George: I totally agree.
Lou: This is a CP framework doc. if PWs removed from framework
will be removed from this doc.
George: There is no such thing as an MPLS-TP-PW.
Andy: MPLS-TP is mpls and the two defined clients of MPLS are IP
and PWs. I expect to see both.
Malcolm: we do need to get this issue resolved in the requirement
document. I would like to see this doc focus on control of MPLS
layer network not what runs above.
George: I will not agree with Malcolm completely. A framework
document has to cover everything and it is only descriptive.
We'll then have protocol specs.
Italo: we need transport and aggregation. The framework
document/reqs should say what is used for what.
Ben: we will draft some text w/ Malcolm to clarify this
20. draft-boutros-mpls-tp-loopback (George)
draft, slides
George [speaking as Chair]: We have lots of OAM related documents>
We'll have to work together and converge on a set that covers the
needed functions.
George [as a co-author]: request is that MEAD takes over all the
drafts presented and sorts out the necessary set of functions and
the set of documents that will describe them.
[Yakov?]: not necessarily end-to-end, how do you do at intermediate
node for two associated LSPs?
George: you need a MIP and a MIP only exists on nodes where the
LSPs go through in both directions
Greg: association may exist in the case of two unidir lsps
George: all this will be cleared-up in framework
Nitin: general comment to oam drafts, why not build onto lsp-ping
framework?
George: We should do as much as we can with LSP-Ping. But I don't
believe it is appropriate for everything.
Rahul: I support doing things with LSP-Ping. We should first try
to re-use lsp-ping and if fails then produce document.
George: If you believe that, write a draft.
Rahul: will do so
Don: suggest your draft should also cover TCM
21a. draft-boutros-mpls-tp-cv (Sami)
draft, slides
[Wei Z.]: need to clarify distinction between MEP/MIP and LER/LSR
Greg: is applicable to p2mp?
Sami: good question, but work still needs to be done
Greg: What about proxy ping?
Sami: as I said, still working
Don: how do you support MEL?
Loa: still open issue in MEAD
Malcolm: lot of stuff need to be clarify in oam-framework need to
separate traceroute from end-to-end connectivity verif
George: may need clarification
[Wei Z.]: MEL level should be part of TLVs, and not LSR id but
MIP/MEP Id
Nitin: any trace req for bidirectional non co-routed?
Loa: we mostly need to get these oam drafts to point to reqs
21b. draft-boutros-mpls-tp-fault (Sami)
draft, slides
[Wei Z.]: MIP can't originate OAM
Loa: take this to the list
Malcolm: Why a message to ingress?
Sami: Ingress needs to know
Malcolm: no
George: We'll take this offline
21c. draft-boutros-mpls-tp-performance (Sami)
draft, slides
Sami presented.
Loa: Will not take any questions. Will go next y1731 and ach-tlv
22. draft-bryant-mpls-tp-ach-tlv (Stewart)
draft, slides
Stewart presented. No question/discussion
23. draft-bhh-mpls-tp-oam-y1731 (Italo)
draft, slides
[Wei Z.]: what makes you believe that we need to support AIS
and RDI?
Italo: we need AIS in transport network. 802.1ag is for
enterprise networks and there AIS is not needed.
Nitin: fine not to preclude other solutions, but make sure not to
have many solutions
Italo: agree
24. draft-ceccarelli-mpls-tp-p2mp-ring (Diego)
draft, slides
Nic: I thought there was already draft on p2mp-lsp recovery
Diego: indeed, we got in contact with authors
25. draft-helvoort-mpls-tp-rosetta-stone (Huub)
draft, slides
no question/discussion
26. draft-martinotti-mpls-tp-interworking (Riccardo)
draft, slides
Greg: again question on PW part of mpls-tp or not?
Andy: head-up we'll be presenting a draft in pwe3 enabling to do
one of scenarios described here. i.e. transparent transport
service.