Minutes IETF102: opsawg
minutes-102-opsawg-00

OPSAWG & Ops Area combined meeting
==================================

Tianran kicked off the OPSAWG part of the meeting
 - Main minutes taker : Joe Clarke (Cisco)
 - Jabber scribe : John Scudder (Juniper)

Update on drafts:
    - NAT YANG Module : Will be proposed for the next telechat (August 3), but
    may get pushed to the subsequent one due to agenda load - TACACS+: Michael
    Wang from Huawei commented that he will be presenting the YANG T+ module in
    NETMOD Friday
      * Also commented that Huawei's implementation is _mostly_ compliant with
      the T+ draft, but there are differences to be discussed on the list *
      Warren Kumari commented that he has written two T+ clients in the past,
      and he believes they were compliant to the draft

Toward a Network Telemetry Framework :  Haoyu Song
  Joe Clarke (as co-chair) : what is the goal of this draft?
    - Get consensus on terminology and differences between telemetry and OAM
    - Do we need new technologies or are we going to work on existing ones
    - Next step is to obtain feedback from the community
  Rüdiger Volk (Deutsche Telekom) : You want to map the landscape of telemetry
    - Claim that a new technology is needed?
    - Learning about the environment one is in is a good thing, but we should
    hold off on taking conclusions until we have the full understanding -
    Separate the problem space and existing technologies from the proposal of
    solutions is premature - Song Feels like slide 4 should be the ultimate goal
  Alex Clemm (Huawei) : what is the scope of this work?
    - Perhaps intent conflates too much with this draft; makes it too broad
    - A larger assurance framework could be too broad

Network Monitoring Protocol (NMP) : Robin Li
 - The presentation ran long, and comments were directed to the list
 - There have already been a number of comments on this work prior to the WG
 presentation

Coordinated Address Space Management (CASM) : Chongfeng Xie
 Chris Morrow (Google) : it sounds like you’re building a database and an API;
 how is that IETF work?
   - How can this database be standardized as people use different things?
   - Will Liu (Huawei) : building a multivendor platform and needs this to be
   standardized
 Ignas Bagdonas (as AD) mentioned the previous [Chicago] BOF that took place
 for CASM
   - IETF is not the place to standardize work such as this
   - Should explore the space of registrars more
 Tianran Zhou (Huawei) : is this the only draft you want to propose?
   - You should build actual models
   - Plan is to submit additional documents in addition to this one

SD-WAN Service Model : Chongfeng Xie
   Adrian Farrel : who are the users of this model?  L23SM users are talking to
   the provider
     - SD-WAN model is intended to be used between the customers and the
     providers
   Robin Li (Huawei) : defining a model for SD-WAN is useful but should this be
   moved to the routing area Ignas Bagdonas (as AD) : the topic is appearing
   here because the ONUG needs YANG expertise
     - most probably this work will move to routing; too early to say where to
     put it.

OpsArea Meeting
===============

Network Failures : Ross Callon and John Scudder
 - Excellent presentation by Ross and John garnering applause at the end
 - Warren Kumari (Google) shared an OSPF pointer story
 - Robin Li (Huawei) suggested a blog to catalog more of these stories
 - Ross : someone should take the time to interview and catalog these failures
 from other long-time internet experts
   * Students should listen to this as it will be important for them to learn
   the history
 - Adrian Farrel : how do we extract the lessons from the story?  This needs to
 be written up so people can extract the lesson from the story - Rüdiger Volk
 (Deutsche Telekom) : some of the problems have been determined through deep
 mathematical work vs. operational experience
   * PhD research may have identified the MED problem ahead of time
   * haven’t done sufficient work on identifying data is passed on or processed
   * vendors should provide sufficient tools to do the hygiene tests
 - Ignas Bagdonas : there is already a paper on the history of the internet and
 its protocols

Coordination of Network Management Evolution : Ignas Bagdonas

 Rob Shakir (Google) : nobody cares that they're using YANG
  - People want to use tools like Puppet and Ansible; Provide running code via
  OSS - These specifications are not required - Instead you need to provide
  code and help users migrate to new paradigms - Focus on solving practical
  problems
 Ignas Bagdonas (as AD) - YANG is not the goal
  - It’s a tool that gives us a means to solve a real problem
 Rob Shakir : impossible to use vendor modules
  - Hard to consume
 Rüdiger Volk models are only a tool
 Rob Shakir wants IETF to reach out more to the OSS community
 Balazs Lenglyel (Ericsson) : modules are just another interface like the CLI
  - Why are they more difficult
 Rob Shakir likes the ease of tools like Ansible that provides [Jinja] templates
  - Easier to get started and has some running code and tooling
 Ignas will provide something more tangible in terms of an experiment by IETF103