IETF-83 Routing Area Open Meeting
=================================

http://rtg.ietf.org 
ADs: Stewart Bryant, Adrian Farrel

Scribe: Deborah Brungard 

Administriva and Area Status
----------------------------
Adrian: Reviewed the Note Well and agenda for today. 

Routing Area Working Group Reports
----------------------------------

BFD (Jeff) - No new RFCs. Covered the new MPLS-TP MIB extensions for BFD. Received response from IEEE so looks
like will get BFD over LAGs as a new charter item. The draft will be ready soon hopefully.

CCAMP (Deborah)  Met two sessions. First on OTN and second session on a new topic with alot of interest,
Flexible Grids.

Forces (Jamal) - Reviewed slides. Plan to redo IETF wide LC on LFB document. Completed 3 outstanding drafts.
Adrian noted more people in the room than previous four meetings. Noted would add SDN, and seemed this
attracted people. Didn't have anything this time, but going forward will.

IDR (John) - Very busy agenda. One new RFC. Vigorous discussion on list on why need AS sets.

ISIS (Chris) - Multi-instance ISIS ready to go forward with. Discussed extending the metrics in the extensions.
Few presentations on SPB and services over. And a draft on TRILL which tracks the changes in TRILL.

KARP (Joel) - Lightly attended. One document went to RFC editor. Another one almost ready for ADs. Started
work on with a new coauthor on key tables which will support the key management. Going well.

L2VPN (Gilles) - Main topic remains ETREE. Had a lively discussion on why two drafts. Also lively debate on
STT and redoing TCP header - go with ADs advice. And TRILL, Data Center Interconnect. Noted not in charter.
(Stuart said not in any charter)
 
L3VPN (Stuart did) - Reviewed slide. Five drafts have been published as RFCs (RFC6513-6517) which is good as they
had been in the RFC Editor's queue for some time (over 2 years in some cases) waiting on normative references
to drafts in other WGs. Had some discussion on the list on closing the working group. Opinions were mixed, although
the leaning was towards keeping the WG open at least until it has cleared the WG documents it currently has on
its plate. Adrian and Stewart are reviewing whether to close or not. Thanks the chairs.
Will announce soon new chairs (shortly).

MANET (Dan) - One document with IESG. New document on security threats. And active discussion on manet-dymo.

MPLS (Loa) - A couple of meetings ago added a new chair. Lots of discussions on 32 bit id for LSRs, and ipv6,
not concluded yet. Plan after Vancouver to have another 8 docs going to IESG. Have a good balance now between
MPLS and MPLS-TP.

OSPF (Acee) -  Two new RFCs (RFC6506 and RFC6549). Four new WG documents. So 7 active WG documents, two close
to being sent to ADs.Two new interesting proposals. Service routing (synergy with Transport Instance), revision
coming. And new draft on virtualization of clusters of routers (TTZ)-need discussion on requirements.
We've had these proposals over the years to scale OSPF by hierarchy. So we'll look at this again.

PCE (Julian)  Will meet after this meeting today. We have a new RFC since Taipei - interlayer requirements. New
working document on stateful PCE. Stateful PCE is a hot topic on our agenda. Also some drafts on optical
extensions. Adrian asked - do you want to comment on your recent email to ccamp list? Julian - yes, we feel
extensions for PCE is in PCE. We feel should follow signaling propoosals, so a bit early for discussion on PCE
(wrt to Flexible Grids).

PIM (Mike) - Two new RFCs since last meeting. Two more in your queue. Continuing to progress PIM spec (bis).
Will be surveying operators to see what they support. New proposal, PIM IGMP-MLD translation. If of interest,
will start. Adrian - it's quite important for the survey for us to get responses.

PWE3 (Andy) - Reviewed slides. No new RFCs. Two drafts in Editor's queue. Four with IESG. Discussed deprecating
Router Alert option for VCCV. Good sense of the room supported. Will confirm on list. Discussed P2MP PW signaling
and protection. This will not progress until we have more progress on P2MP PW requirements and architecture.
Also need P2MP PW OAM. Discussed capabilities to explicitly map a PW to a PSN tunnel. Very spirited discussion
on PW endpoint failure detection.

ROLL (JP) - We reached a milestone to publish 3 RFCs. Biggest milestone since formed the group. Need to spend
more time on security aspects. Mike Richardson is new co-chair. Start to see deployments of ripple,
had some feed back with one which was quite large. Encourage to come to group and tell about deployments.

RTGWG (Alvaro) - Will meet, have 9 presentations. Since previous meeting, completed LFA work.
Started new work on manageability. Published an update
to an IP Fast Reroute MIBs draft. And work on operational manageability.

SIDR (Sandy) - Met twice to very full agendas. Monday, long discussion on route leaks. CPU loads, interesting
discussion on interpretation of validation. And reports on performance aspects. Wednesday good discussion on the
protocol. A presentation on threats, and one on interoperability. And one on updates without slides (which got
a compliant). One on keys for routers. And one on DNS which had quite a bit of discussion. Stewart - interims
planned? Sandy - will have monthly interims between now and Vancouver. Probably along side the RIPE meeting.
Still working on dates.

NVO3 BOF (Matthew) - Had over 250 people in BOF, concentrated on requirements and architecture. Have deliverables
around that and also to do a gap analysis. Had general agreement in the room on doing this work and the charter.
Stewart - I've requested to the two BOF chairs to have the charter to the IESG for Friday and will request a
slot on the telechat. I strongly support it.


Lightning Talks
----------------
Adrian presented slide on objectives of this for meeting - objective is to educate on other areas and
recommend that you take any comments to those groups. Not intention here to debate - go to the working group.

HOMENET (Acee)
Reviewed slides. Adrian had asked for me to be advisor for this group. So I've been attending. Scope is to
develop an architecture for IPv6 home networks.
Most important thing right now is to agree on architecture document which has requirements. Prefer to use
solutions which already exist and how would be deployed.
Has to be lightweight.Reviewed slide with requirements. Convergence time of minutes is ok. Acee thinks a bit
loose but they didn't want to disqualify e.g. RIP. Want to be autoconfiguring, plug and play. Direction so
can maintain routing w/o upstream ISP connectivity. Most important thing right now is to get comments in on
these requirements.
A number of canidates for routing protocols.Some think the link state protocols may be a bit heavy. We've
pointed them to some lighter implementations of OSPF. Strong lobby on Ripple as then would automatically
have support for low powerlousy networks (LLN). Also looking at Babel (open source project).
Jeff- I've been watching the mailing list, what sort of provisions for users to choose for a network e.g
can only get vpn to work with AT&T.
Acee - source based routing.
Jeff - may not be enough - policy based routing based on a tuple.
Acee - maybe part of auto configure?
Jeff - as home network should be easy to use. And source routing may add to cost.
Acee - for source routing with possible support by Linux may be cheaper.
lou - maybe should ask reqs for home networks - how many are dual homed really among regular users?
adrian - probaly a homenet discussion - so shouldn't really meant to dissect it now.
ray - homenet chair - yes appreciate if do on our list. we still have a lot undecided. Fred baker has a
draft on policy based routing. Please come and participate.

CDNi (Stefano)
Reviewed slides. Routing technologies for non-routing technologies. CDNI (Content Delivery Networks Interconnect)
is a CDN mesh/federation. Needs to decide where the user should get the content?
Multiple proposals. One makes use of MP-BGP (with extensions). This could be used if AS reachability
information is enough.
The BGP advertisement may not be enough - may need new attributes which we have identified for BGP. For
CDN, not just footprint and reachability, also need to use business aspects.
This runs as overlay over the infrastructure. Will run in CDN routers - not routers or the routing layer.
Passively receives BGP database. Other examples include BGP-LS, ALTO.
Reviewed concerns and issues. Feels should relax a bit the rules of what can carry in routing as this will be
used outside of the routing layer.
Noted reusing does not mean fate sharing with the routing layer. Recognize need to develop deployment
guidelines to ensure this.
Acee - I haven't read any of this. You talked about prefixs - these are not prefixes - they are handles how to
hand off?
Steffano- These are.
Sue - What you want to hear is how to keep these separate from BGP routing (that's what I heard on Monday
from you) I'm hearing a little bit different now.
Steffano - yes, will not be in routers.
Acee - so BGP or not to BGP?
(laugh)

Reload P2P SIP (Cullen)
Cullen - Thought wanted about SIP routing. Didn't include Reload P2P.
Adrian - that's ok.
Cullen- Reviewed slides. This is leading towards some problems we had. A new Working Group called that's
forming. We have email style names. But no one uses them - they use the telephone addresses. But unfortunately
in deployments we use something a bit different so a fuzzyness.
We had extended BGP for distributing numbers (TRIP) but no one is really using. Most are managed with Excel
spreadsheets which limit how fast can update.
Biggest problem is with Loop Detection (slide). As it's hard to differentiate a spiral (which is ok) from a
real loop.
Because of topology hiding, it is difficult to do loop detection. Can have infinite loops.
Would welcome ideas on how to pass information without breaking.
Acee - are these at arbituary points or a discovery for within provider or between providers?
Cullen - at edge nodes will know whats in, and they will have configuration on what other service providers
connect with.
Adrian - the trick which was done was to use a different port than BGP?
Cullen - yes, use a different port, but can really reuse most of BGP.
Stewart - can come back and talk about reload sometime?
Cullen - be glad to.

DHCPv6 (Wojciech Dec)
Reviewed slides. Basic scenario is a multi-homed client. It's about configuring static routes in a convenient
manner. Not suitable for networks that do dynamic routing. Key problems being addressed: deal with cases of
multiple interfaces, ability to configure individual hosts on multi-host segments, difficulty of managing per
host configuration on each edge router.
(No questions)


Information exchange between Routers (Dimitri)
----------------------------------------------
Reviewed slides. If agents run on "routers" then the routing system becomes a multi-agent environment. Consequently,
we need to enable exchange of information between agents to perform various functions. The question is if the
IETF is the place to specify this protocol. Planning to have a BOF in Vancouver.
Stewart - on the BOF - in which area?
Dimitri - I was thinking in the RA. if look at this in perspective of all our discussion e.g. extends BGP,
this is not the right question. It should be - do we have the right technologies?
Adrian - I wouldn't be against if a body of people would come to talk about it, but for a BOF in Vancouver
need to start spinning the wheels.
Adrian - for the room was this useful?
Room - yes.


Open Discussion
---------------
No discussion.