Skip to main content

Minutes IETF99: isis
minutes-99-isis-00

Meeting Minutes IS-IS for IP Internets (isis) WG
Date and time 2017-07-17 11:30
Title Minutes IETF99: isis
State Active
Other versions plain text
Last updated 2017-08-02

minutes-99-isis-00
IS-IS Etherpad
Scribe: Acee Lindem (acee@lindem.com)


  IS-IS WG Agenda IETF-99

Time Slot (120m): Monday, July 17, 2017 13:30-15:30 CEST

* Intro, Adminastriva, Document Status
        Presenter: Chairs (Christian Hopps, Hannes Gredler)
    - OpenFabric will not be presented.

* Combined IGP-LS GROUP Proposal - Chris Hopps
    - Lots of overlapp
    - However, audiences are not exactly the same.
    - May be harder to reach consensus.
    - Proposal: Combine OSPF an IS-IS
        * Run join session in Singapore.
        * Combne mailing list
  Loa Andersson: Doesn't see it as necessarily less work.
  Chris Hopps: Could potentially have one draft with OSPF
               and IS-IS.
  Loa: Would be very confusing with one draft.
  Les Ginsberg: Could start with back-to-back sessions in
                one meeting.
  David Lamparter: Believe protocol user space is not that
                   diverse.
  Chris: Believes smaller operagtors don't attend IETF.
  Chris Bowers: Thinks that common use case documents would
                be useful.
  Mikal Abrahamsson: Do we need effort to align features between
                     protocols?
  Acee Lindem: Believes we have the features available to do
               anything we want in support of features. There
               will be different protocol mechanism
               enhancements.
  Hannes Gredler: Room poll supporting merger - about 50/50
                  for/against.


* IS-IS TE Attributes per application
        Presenter: Les Ginsberg
        Document: https://datatracker.ietf.org/doc/draft-ginsberg-isis-te-app/
  Chris Hopps: Why do you need legacy bit?
  Les Ginsberg: How do say not to use the new attributes?
  Peter Psenak: This gives you explicit advertisement for all applications.
  Les Ginsberg: Gives you a set of attributes for an application.
  Hannes Gredler: Deviation from how we have migrated IS-IS in the past?
                  How do support old and new?
  Les: This is how to use attributes for new applications - not using
       different attributes.

  Bruno Decraene: How does this work with attributes used for remote-LFA?
  Les: You can use legacy advertisements with partial deployment or
       use the new attributes for full deployment.


* Advertising TE protocols in IS-IS
      Presenter: Chris Bowers
      Document: https://datatracker.ietf.org/doc/draft-hegde-isis-advertising-te-protocols
   Raveendra Toriv: Is this multi-topology?
   Chris Bowers: SR and TE topologies are congruent. This is not multi-topology.
   Les: Hope we agree we are not modifying base SPF.
   Chris Bowers: Yes
   Harish Sitaraman: Why aren't admin groups a good long term solution?
   Chris Bowers: Requires configuration and prone to errors.
                 For example, a new link where you want to advertise
                 bandwidth.
   Hannes: Your draft assumes congruent topologies?
   Chris Bowers: Next draft addresses different attributes per application
                 with different topologies.

* Ext. to IS-IS to Assoc. TE Attr. with TE Attr. Sets and SRLGs with SRLG Sets
   Presenter: Chris Bowers
   Document: https://datatracker.ietf.org/doc/draft-bowers-isis-te-attribute-set/

* Resolve The Conflict
   Chris Hopps: Seems like encodings are different but the problems solved
                are the same. Can IS-IS TE apps draft be used as a base
                for consensus?
   Chris Bowers: Difference in requirements.
   Les: No way to say not to use legacy attributes. Do not have any standard
        bit assignments. It is local configuration so how does the
        controller know the mapping?
   Chris Bowers: Do you believe the first TE enablement draft can be adapted?
   Les: Not needed.
   Tony Przygienda: Need to agree on the set of requirements.
   Hannes Gredler: Can it be done using existing protocols? This could be
                   done using multi-topology?
   Les: Are you suggesting we use topologies for applications?
   Hannes: Yes
   Les: We are both scoping attributes in different ways.
   Hannes: But why not existing multi-topology?
   Acee: OSPF will not use multi-topology due to lack of deployment.
   Tony Przygienda: Have used multi-topology for other problems.
   Ron Bonica: Maybe we should have an interim on use cases and
               requirements.
   Julien Meuric: Could allocate multi-topology code points for
                  various TE flavors.

* IS-IS Routing with Reverse Metric
        Presenter: Mikael Abrahamsson
        Document: https://datatracker.ietf.org/doc/draft-ietf-isis-reverse-metric
   Chris Hopps: If you have crappy signal on your fiber, you can advertise a
        high reverse metric. This will greatly discourage usage but it can be
        still used in the other direction.
   Chris: Can we go to WG last call with this document? No objections
          Think it is good idea? Support but not enthusiastic.

* IS-IS Routing for Spine-Leaf Topology
        Presenter: Les Ginsberg
        Document: https://datatracker.ietf.org/doc/draft-shen-isis-spine-leaf-ext/
   Chris Hopps: Cherry picked Russ' ideas from OpenFabric?
   Les: Yes, but with his permission.

* ISIS Extensions for Flexible Ethernet
        Presenter: Mach Chen
        Document: https://datatracker.ietf.org/doc/draft-zcdc-isis-flexe-extention/
   Julien Meuric: Discuss GMPLS extensions should be in TEAS and CCAMP. Need to
                  clearly define usecases for Flex Etherent and GMPLS.

Ended exactly on time.