Dear all,
please find attached the minutes from the ALTO session at IETF99.
Please let Vijay and myself have any comments/corrections within a week.
Many thanks to our note-takers Lyle and Sabine!
- Jan
***
IETF 99, July 16-21, 2017. Prague, Czech Republic
ALTO WG Meeting, Thu Jul 20, 2017 1330-1530 Karlin III
Notetakers: Lyle Bertz, Sabine Randriamasy
1330 - 1340 10' Agenda bash and WG update (Chairs)
- Focus is to finish chartered items so the agenda is based on this
(charter & milestone items are first; new topics are best effort)
- Goal is to finish milestones @ IETF-101
- Richard : Finish means?
Jan: If something is close to WGLC that should work; in
principal finish means WGLC
- Last meeting: adopted path vector, unified prop, cdni
- milestones: Multi-Cost OK
- ALTO performance metrics needs more work (review)
- Props, Path Vector & CDNI need more work
- Other business?
- Richard:
- What is the timing for the next wg meeting if not meeting @
IETF 100?
- Jan: Timeframe is around Singapore
- AD: Clarification that meeting should not be at the time of
IETF but closer to IETF 100
- Workshop is mid August in Los Angeles with LHC and other teams.
1340 - 1355 15' ALTO cost calendar, S. Randriamasy
https://tools.ietf.org/html/draft-ietf-alto-cost-calendar-02
- Reviewed all updates
- Discussion Items
- - RY several time granularities. Client can aggregate itself ?
=> SR: yes the server should provide finest granularity.
(TODO: the client SHOULD interpret the time interval
size as the finest available from server).
- Richard / Sabine agree to move ahead on granularity response
- Sabine - The ALTO server provides the finest grain value for
the time-interval
- Jan: After issues herein are closed does the team agree this should
WGLC (hum)
- Result: No hums against moving it to WGLC once issues/open
items are resolved
- Please make the resolutions on this on the list
1355 - 1410 15' ALTO cost context, S. Randriamasy
https://tools.ietf.org/html/draft-randriamasy-alto-cost-context-02
- Reviewed Concepts and Updates per WG feedback
- Questions:
- Dawn - One suggestion is to make suggestion to specifiy the
order of the response (has this been considered)?
- Different options were discussed. Will continute to mailing
list.
- Would adding context to the cost add extra privacy concerns?
- This should be discussed in the document.
- Richard - point out that this is part of a more generalized
issue of a type system in ALTO
- Type system was urged to be a future dicussion on this matter
- Richard: we need reconsider cost type design (type, mode,
where fit context?)
- Seb: IANA considerations ?
==> SR yes we need look at other IETF Work and consider
IANA registration.
- For now, chair ask is to move ahead but since it is not part of the
milestones we will lead
1410 - 1425 15' ALTO incremental updates, R. Yang
https://datatracker.ietf.org/doc/draft-ietf-alto-incr-update-sse/
- Overview of changes
- Does WG believe this is ready for WGLC after items are completed (hum)?
- No reaction either way
- Sabine - Main task is complete but needs work to specify who is
sending a SSE and for what reason
- Suggestion is that we can make the Client/Server roles more
explicit
- JS: JSON patch important for CDNI. WG seems now more active on that
one.
- SR: needs clarification on who requires what for what
- RY: yes we will need to clarify that
1425 - 1435 10' CDN Footprints and Capability Advertisements, R. Yang
https://www.ietf.org/id/draft-seedorf-cdni-request-routing-alto-10.txt
(This is a WG draft; needs to be resubmitted with the correct
name.)
- Jan presenting
- Advanced design is being looked at but the focus is the basic design
- Richard - Discussion wrt the envelope name
- outcome is to use somehting like "cdni/fci"
1435 - 1450 15' ALTO path vector, D. Chan
https://datatracker.ietf.org/doc/draft-ietf-alto-path-vector/
- Discussed 3 remaining design issues
- Point raised about why this technique for MIME is not used in SSE
- Privacy implications exist for cross product query but PID flows don't
1450 - 1455 05' Extensible property maps for ALTO, D. Chan
https://datatracker.ietf.org/doc/draft-roome-alto-unified-props-new/
(This is a WG draft; needs to be resubmitted with the correct
name.)
- With no time left, discussion will be deferred to the list
- Jan: Please don't add use cases every 2 months or it will never get
done
1455 - 1500 05' ALTO cross-domain server discovery, S. Kiesel
https://datatracker.ietf.org/doc/draft-ietf-alto-xdom-disc/
- Richard - Document is useful but if we can carve out the principal
it would be helpful, can you come up with more clear principles?
- JS: seb said will look at it.
- Will review Richard's comments and move to WGLC
1500 - 1509 09' ALTO cellular address extension, S. Randriamasy
https://tools.ietf.org/html/draft-randriamasy-alto-cellular-adresses-00
- Proposed based on prior work
- Lyle - Ignore leading 0 concern as they are about LR(0) parsing and
over the wire frame alignment
1509 - 1518 09' ALTO extension: Flow-based cost query, J. Zhang
https://datatracker.ietf.org/doc/draft-gao-alto-fcs/
- No time for discussion. All comments/discussions pushed to the
mailing list.
1518 - 1527 09' Resource Orchestration for Multi-Domain Data
Analytics, Q. Xi
https://datatracker.ietf.org/doc/draft-xiang-alto-exascale-network-optimization/
(Possible remote presentation)
- No time for discussion. All comments/discussions pushed to the
mailing list.
1527 - 1530 03' Wrap up / Other business
- Next virtual interim meeting will be posted on the mailing list
AO = ALTO
AOS = ALTO Server
AOC = ALTO Client
***
_______________________________________________
alto mailing list
alto@ietf.org
https://www.ietf.org/mailman/listinfo/alto