Skip to main content

Minutes interim-2017-alto-01: Mon 06:00
minutes-interim-2017-alto-01-201712180600-00

Meeting Minutes Application-Layer Traffic Optimization (alto) WG
Date and time 2017-12-18 12:00
Title Minutes interim-2017-alto-01: Mon 06:00
State Active
Other versions plain text
Last updated 2017-12-19

minutes-interim-2017-alto-01-201712180600-00
ALTO Virtual Meeting
Monday, Dec-18-2017 6:00am US Central - 8:00am US Central
Notes taken by Jensen Zhang and Y. Jace Liu
Jabber monitored by Richard Yang

Meeting materials at
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/

Attendees: 13 (including chair).

Chair Slides and WG Status, Vijay K. Gurbani; slides at [1]
------------------------------------------------------------
Agenda bashed.  Last f2f meeting was in Prague (Jul 2017).  Plan to meet
in London (Mar 2018).  Milestones out of date; chairs to coordinate with
AD.  Multi-cost RFC 8189 published in October 2017.

Vijay went through the WG deliverables and status of each deliverable.
Some of the I-Ds have agenda time during the meeting, so we will get an
update during the presentation.  For other I-Ds assigned to deliverables,
Vijay noted the following:

- Action item for xdom-disc authors: Please see slide 7 [1] and post an
update to the mailing list on the status of this document.  The WG needs
to move this work ahead as the impression is that it is mature.

- Action item for incr-updates-sse authors: This I-D fulfils two milestones
(slide 9 of [1]).  Authors to post quick update on status of this I-D on
mailing list.
Richard Yang (as author): I-D needs some minor changes that will be completed
after the end of year break.

- Action item for performance-metrics authors: Please see slide 10 [1] and
post an update to the mailing list on the status of this document.

CDNI Request Routing, Shawn Lin; slides at [2]
----------------------------------------------
Shawn went through some changes in the draft.  There do not appear to be any
major open issues with the draft, although Shawn wants to add more content.
Chair suggested to revise as soon as possible and post on mailing list to
move the work ahead.  Richard Yang agrees and will plan a sync-up meeting
among the authors.

ALTO Cost Calendar, Sabine Randriamasy; slides at [3]
-----------------------------------------------------
This work is fairly mature and may suffer from feature creep if we are not
careful (see slide 10 of [3]).  The general discussion of the working group
was that we should close editorial issues with this draft and move ahead to
WGLC instead of adding new functionality in this draft.

- Action item for the chairs: Ratify the decision to go to WGLC on list and
move work ahead.  The work has been reviewed by Jensen and Dawn.

PANRG Lightning talk on ALTO, Sabine Randriamasy; slides at [4]
--------------------------------------------------------------
Sabine presented ALTO as a protocol for PANRG that may provide PANRG a
ready-made solution for dealing with path properties and endpoint selection
of paths.  PANRG appears to be aware of ALTO and ALTO is presented as a
candidate protocol along with SCION (slide 4 of [4]).

- Action item for Sabine and Dawn is to continue fostering the discussion
with PANRG, and if PANRG would like to make a f2f presentation in the ALTO
WG on their requirements during the London IETF, that may be beneficial.

ALTO Path Vector, Dawn Chen; slides at [5]
------------------------------------------
The working group felt that the work is now mature; Sabine has reviewed this
work extensively.  Considering the importance of this draft to the re-
chartered ALTO WG, the chairs would like to have an additional sustained
review of the work.  Chairs to solicit volunteers for a second review.  There
are no open issues that the authors are aware of besides ensuring that SSE
and path-vector work well with each other (i.e., no implicit dependencies).

- Action item for chairs to solicit an additional reviewer for this work.

Unified Property Map, Dawn Chen; slides at [6]
----------------------------------------------
There is a dependency between unified-props-new and path-vector; latter
depends on the former.  Thus unified-props-new needs to proceed first, or
both drafts can proceed as a bundle.

Some discussion on representing cellular addresses in ALTO.  Currently,
cellular addresse format is specified in a companion draft (individual,
draft-randriamasy-alto-cellular-ssresses-01).  Decision for the working
group is two-fold:

(1) Are cellular addresses an important abstraction that the working group
will like to maintain in ALTO?
(2) If yes, is the unified-props-new draft the correct place to add the
cellular representation?

- Action item: Chairs to post these questions on the mailing list.

Exascale network optimization, Qiao Xiang; slides at [7]
--------------------------------------------------------
This is an individual draft; however, the chair (Vijay) and the working group
(Richard Yang, Sabine Randriamasy) feel that the work is important enough to
archive.  It is an excellent use case of how ALTO can help with resource
orchestration in large-scale networks.  Possibilities are to move this as an
Informational RFC either sponsored by the WG or by the AD.

- Action item: Chairs to determine how to effectively move this work ahead.

ALTO Flow-based Query Cost Extension, Jensen Zhang; slides at [8]
-----------------------------------------------------------------
This is an individual draft as well.  Richard feels that the flow extension
problem is an important one to tackle in ALTO.  However, the working group
needs to decide how to move this work ahead.

- Action item: Chairs to determine how to effectively move this work ahead.

ALTO Path Vector Compression, Kai Guo; slides at [9]
----------------------------------------------------
This is an individual draft that depends on the path-vector work.  As with
the other individual drafts, need to decide what to do with this work.  The
first order of business is to finish the current deliverables and milestones.

References:
[1]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-chair-slides/
[2]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-cdni-request-routing/
[3]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-cost-calendar/
[4]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-panrg-lightning-talk/
[5]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-path-vector/
[6]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-unified-properties/
[7]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-resource-orchestration-for-large-scale-multi-domain-data-analytics/
[8]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-flow-based-cost-query/
[9]
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-routing-state-abstraction/

--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
Raw notes from Jensen Zhang and Y. Jace Liu

IETF 100, ALTO WG Interim Meeting, Mon Dec 18, 2017 0700-0900, US-ET

Audience: 16 people on Webex
Meeting note taker: Jensen Zhang, Y. Jace Liu
Jabber scribe: Y. Richard Yang

=== Chair Slides, Vijay K. Gurbani ===

Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-chair-slides/

Vijay confirmed any changes of the schedule and Richard added a slot of SSE
incremental update. Vijay thought there are many things to move forward
massively, and introduced every project progress. CDNI FCI: the author was in
the meeting, and there is existing a draft. The author would move ahead.
Alternative Server Discovery Document: author was not online, and Vijay would
ask the author to send updates of the draft on the mailing list. Path vector:
the author was in the meeting, and there is existing a draft. The author would
move ahead. Endpoint property extension: the author was in the meeting, and
there is existing a draft. The author would move ahead. Server-initiated
notifications and incremental update: the author was in the meeting, and there
is existing a draft. The author would move ahead. Richard said some minor edits
would be completed right after the year's break. cost property extension
document: alto cost calendar would took one slot of the meeting, and the author
would move forward the updates of the draft. The author would send updates of
performance metrics on mailing list.

=== CDNI request routing, Shawn Lin ===

Draft:
https://datatracker.ietf.org/doc/draft-ietf-alto-cdni-request-routing-alto/
Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-cdni-request-routing/

There are some changes in the draft. 1, The name of CDCI FCI ALTO Service has
been renamed to CDNI FCI Map Service, to follow the name convention. 2, The
draft suggested to use JSON Patch for incremental update. 3, Some examples have
been added in the draft. 4, The draft include new filters on capabilities and
footprints.

Vijay asks for the open issues. Shawn said that he want to add more content
into the ietf draft. Vijay suggested to revise it as soon as possible so that
people could look at it and move to the next step.

Richard said Shawn has made some differences and is trying to get feedbacks
from Kevin, Jon and Jan. They would plan a sync-up meeting to get feedbacks,
then they would post remaining issues and updates on the mailing list.

=== ALTO cost calendar, Sabine Randriamasy ===

Draft: https://datatracker.ietf.org/doc/draft-ietf-alto-cost-calendar/
Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-cost-calendar/

Based on the working group feedbacks, some changes has been made to the draft.
Some design about calendar attributes was updated, and some clarifications is
included in the draft. Besides, the draft was reworded and cleaned up.

Vijay gave his feedback that the project is relatively mature, and should not
include new stuff there. Richard held the second opinion, the new stuff and
design could be contained in next version, but it’s better for now to focus on
the wording problems. Dawn agree with the idea that the calendar should move to
working group last call and nobody object to it.

Decision: calendar should move to working group last call.

=== PANRG lightning talk on ALTO, Sabine Randiamasy ===

Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-panrg-lightning-talk/

Vijay thanks for the presenting of ALTO Lightning on PANRG, hoped Dawn and
Sabine could follow up on the London IETF meeting, to see if anything proactive
we can do. Richard agree and demonstrate that the great use cases really match
the topic we’ve done and we are interested. Because of the location of members
of PANRG group, they can’t attend the meeting, but Dawn and Sabine have done a
great job to have conversations with them.

Sabine said It would be great if PANRG could get a slot in the meeting, to talk
about how they look at the material. Vijay and Richard agreed.

Qiao noticed that on the of the key question of PANRG is “trustworthy” of path
properties, and he was wondering what the “trustworthy” is about. Qiao has also
concerned if ALTO should provide information about the “trustworthy”. Vijay:
please explain briefly. Sabine: The only thing I mentioned related to that is
in the security consideration of the base protocol. Typically because it’s the
information exchange between operator and hosts that is subscribing to the
operator, there has to be an assumption of trust between the 2 parties, so this
is the only way I agree. There is no mechanism that ensures mutual trust. It’s
kind of a mutual incentive to ask for reliable requests and responses. Qiao:we
could take a discussion offline.

=== ALTO path vector, Dawn Chen ===

Draft: https://datatracker.ietf.org/doc/draft-ietf-alto-path-vector/
Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-path-vector/

Dawn discussed a lot of updates on the path vector based on the review.

Vijay: Thanks for Sabine’s recent review. We need one more review, because this
is an important output as a recharter. Anybody want to be volunteer, please
send me and Jan an email. Vijay: Any open issues from Richard? Richard: Only
thing would be to make sure the multipart would work well with SSE. We need to
review it carefully. Vijay: It is an important point. Call for volunteer. Will
talk with Jan about this.

=== Unified property map, Dawn Chen ===

Draft: https://datatracker.ietf.org/doc/draft-ietf-alto-unified-props-new/
Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-unified-properties/

This revision only does some minor changes.

Richard: Path vector depends on the unified property. We need to proceed the
unified property before the path vector. Sabine: Register the cellular address
as a domain in the unified property draft? Dawn: You can register it in the
individual draft Kai: Could it introduce some draft dependency issues Vijay
proposed two questions to discuss later:

A: Does WG feel the cellular addresses are the important abstraction?
B: If they are, can they be put in the unified property?

=== Exascale network optimization, Qiao Xiang ===

Draft:
https://datatracker.ietf.org/doc/draft-xiang-alto-exascale-network-optimization/
Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-resource-orchestration-for-large-scale-multi-domain-data-analytics/

Qiao talked about the recent deployment experience on the multi-domain
scientific data analytics use case.

Vijay pointed this was a very good work. Need to talk with Jan how to move the
draft ahead. Richard commented that it would be nice if this work could be an
informational document. Because it is a pretty good use case. Vijay agreed on
this.

Sabine: This work is about the multi-domain science data analytics system. Is
it possible to extend the work to other types of analytics systems. e.g. the
IoT system. Qiao: Good point. Will discuss it offline.

=== ALTO flow-based query cost extension, Jensen Zhang ===

Draft: https://datatracker.ietf.org/doc/draft-gao-alto-fcs/
Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-flow-based-cost-query/

Jensen reviewed the updates and FCS additional requirements on ALTO. The
current solution for the validation problem is discussed.

Vijay: It is still an individual draft. We need to decide how to proceed it.
Richard: The flow extension problem is important. And the service should be
valuable for ALTO. If we can finish this draft quickly, we can have chance.
Sabine would like to write some comments and post them in the mailing list.
Sabine mentioned that the idea of avoiding full-mesh request maybe more
important than introducing the port number.

=== Compression ALTO path vectors, Kai Gao ===

Draft:
https://datatracker.ietf.org/doc/draft-gao-alto-routing-state-abstraction/
Slide:
https://datatracker.ietf.org/meeting/interim-2017-alto-01/materials/slides-interim-2017-alto-01-sessa-alto-routing-state-abstraction/

Routing State Abstraction (RSA) proposed a reference implementation of the path
vector. The major change of this version is to cleanup the routing state
abstraction algorithm. Kai gave a brief introduction.

Vijay: This draft depends on the path vector draft, or opposite?
Kai: It depends on the path vector.
Vijay: Like flow query, we need to open some discussions to see whether people
in WG are interested in it. Richard: That is a good idea. It will be related to
recharter.