Minutes IETF101: opsawg
minutes-101-opsawg-03

What: Combined OpsAWG / OpsArea
When: 13:30-15:30	Monday Afternoon session I
Where: Blenheim

Ops-Area Section
---------------------	
Administrivia - scribes, minutes, etc.
Benoit / Warren / Ignas
10 minutes	

AD transition

Warren:
Much thanks to Benoit, and a bottle!

Benoit - Many YANG Modules 3214 in the catalog as of this morning.
What needs to be improved: 
 - how do we update the modules efficiently ?
 + need major and minor version control.
 + great to have RFC number, but need lighter update process
 + need EXAMPLES, and auto-updates for them.
 + IANA validation of the latest version?

Eliot Lear - thanks for being the *level head in the room*
Issue with Coordinator:
Problem with YANG isthat there is code associated.
Need to commitment to make updates from new model champions,
 to achieve Operational Excellence!
 
Ignas - Develop the models that address the actual OPS need.
need to focus on the requirements FROM USERs (for models, etc.)
need to evolve our work to the useablility phase.

Joe Clarke:
How do we move faster, more like the OS model?  The goal is:
Be better about producing OPS need at the required velocity!


Open Mic
No questions!

Ops-Area Working Group Section
------------------------------
Administrivia - scribes, note well, etc.
Ignas / Tianran / Joe

Ignas is transitioning to become Ops Area AD after Benoit.

Joe has become a third co-chair in opsawg and after Ignas becomes AD, opsawg will
continue to have two co-chairs for the time being

Tianran provided an update of drafts:
 + draft-ietf-opsawg-capwap-alt-tunnel-12
  - In RFC Editor Queue
 + draft-ietf-opsawg-ipfix-bgp-community-05
  - First Last Call
  - Could benefit from more operator review
 + draft-ietf-opsawg-nat-yang-13 and draft-ietf-opsawg-mud-18
  - Sent to IESG and pending telechat scheduling
 + draft-ietf-opsawg-tacacs-08
  - In progress (see below)
  
Alan De Kok came to the mic to discuss draft-ietf-opsawg-tacacs-08
 + Authors are not engaging with the working group
 + Periodically, a revision that addresses some comments is submitted with no follow-up
   discussion
 + Authors have used verbatim Security Considerations text without attribution
 
Joe agreed with Alan on the engagement and took the AI to follow up with the authors
 + Since then, revision -09 has been posted addressing some of Joe's contributor reviews
 + Authors committing to a new revision with follow-up to address Alan's remaining issues
 
Eliot Lear presented a MUD update
 + Current document awaiting April 19, 2018 telechat
 + Some minor changes will be required prior to ratification
 + Next round of work will include how to describe bandwidth usage
 + While not strictly IETF-focused, 802.11 on-boarding is a prescient issue for IoT devices
 + URL composition and domain name lifetimes also need some thought
 
Zhenqiang Li presented draft-ietf-opsawg-ipfix-bgp-community
 + All comments and suggestions incorporated in the draft
 + BGP community information can now be exported with specific flows
 + He asked if it's ready for the next step
  - Benoit commented that this draft should not say it "Updates" RFC7012
  - Four people have read revision -05
  - New version will be published that removes the "Updates" text
  
Haoyu Song presented draft-song-ntf : Toward a network telemetry framework
 + Visibility is necessary for control
 + More visibility is required
 + Existing OAM framework is insufficient for Intent-Based Networks
 + Lots of piecemeal solutions exist
 + Lack of collaboration
 + Data quantity will be large and must be processed in real-time
 + Data comes from many different sources
 + The network needs to pre-process the data
 + But measurement functions must not interfere with the main function of the devices
 + Telemetry should partitioned into different "planes"
 + There should be a unified solution space
 + Ignas commented (as a contributor):
  - "I get the feeling you are trying to boil multiple oceans."
  - At the same time the majority of components already exist; areas already covered
  - This feels more like product design rather than protocol design
  - What is wrong with the current OAM?
 + Authors reiterated the need for a unifying framework
 + Tim Carey stated, intent could be realized in many different ways and each with many 
   sources of data and from different protocols. How to correlate those sources of data 
   back to intents would be highly valuable.
 + Laurent Ciavaglia commented some functions do not seem to be addressed in the document,
   and asked for clarification on the data transport/distribution, as well as data storage
   functions and their respective optimization.
 + Joe commented (as co-chair) that more scoping is needed for this work
  - Perhaps focus should be specific to Intent-based networking
  - Currently, it feels too broad, especially given that it touches so much work already
    in progress or that has been ratified
 + Authors will look to refocus the text
 + More reviews are needed
 
Chongfeng Xie presented draft-li-opsawg-address-pool-management-arch : Coordinated Address
Space Management
 + CASM is a general architecture to handle the management of address pools
 + This document can form the basis for future work on modeling and design
 + Use cases include:
  - Address poll configuration
  - NAT & CGN
  - Address configuration of IPAM
  - SDN controller
  - Interfaces to the RPKI
 + Field trial of a CASM solution conducted between Huawei, ZTE, H3C, and Certus
 + Joe asked what has changed between the BOF held in IETF98 and now
  - There were comments that came out of the BOF that were merged into the text
  - Two drafts that were worked on prior to the BOF have been merged
 + Joe as a contributor
  - Looks like two fused drafts
  - There was confusing terminology
  - The text currently references expired drafts
  - Problem space is interesting
  - More refinement is required for the draft
 + Tianran polled the WG for interest
  - Six have read it
  - A few more show interest
 + More discussion is needed on the mailing list
 + Ignas had a conversation with enterprise operators
  - Lots of interest in this area
  - ONUG consists of a number of enterprise operators that have a need for this type of solution
  - Solution should be extended to handle containers, VLAN assignments, etc.
  
Qin Wu draft-wu-opsawg-network-overlay-resource-model
 + Define a resource-facing model for resource management (network re-optimization, 
   path selection, binding between the service and the path 
 + Bridge between Service-centric models and network-centric models
 + This resource model can be mapped from the Customer-facing service model
 + Out of time in the meeting
 + Joe asked how many people have read this
  - 10 have read it
  - Seems to be lots of interest
  
One last round of applause for Benoit!