On Monday, Jul 4, 2016 "Y. Richard Yang" <[email protected]> wrote:

Vijay, Jan, all,

I am replying this email publicly so that all of us can engage more
in shaping the agenda in about three weeks.

A key item that the WG needs to discuss, in the potentially more
productive f2f-meeting setting in Berlin, is how the WG should move
forward. Can we allocate a slot, say at least 20 min, for this
discussion?

Richard: ACK.  More inline.

Besides the general discussions, here are some comments on specific
agenda items and the mission of the WG:

- Despite the substantial change of the networking landscape, I still
believe that the foundational services of ALTO are sound: providing
abstract network information to applications. I believe that this is
also what many others see ALTO; for example see [1]. The key is what
we do to go beyond the initial network abstractions.
>
- The first I see is Endpoint Cost Service (ECS), which is the
foundation. It provides an interface to allow applications to know
the routing costs. However, current ALTO is defined pre-SDN, and the
emerging trend of SDN is going beyond simple destination routing, to
be more application dependent. Hence, I see that we need to go beyond
Endpoint Cost Service, to say Flow Cost Service. There is an
extension draft [2], and I suggest that the WG finishes this to make
the ALTO protocol complete. Make sense?

The ecs-flow draft has been around for a couple of months.  I don't
think it has attracted WG attention, despite being headed in the right
direction with respect to the abstraction that involves application-
constraints during routing decisions.

-  The next I see is network graph (path vector) abstraction. The
more I think about it, the more I am convinced that this is the way
we go. I am an author of the draft and multiple versions are evolving
[3, 4, 5], but we are relatively slow in convergence. I suggest that
the WG gets a set of active participants to finish this draft, so
that we have a relatively complete set of ALTO services.

The above will be imperative, I believe.  I may well be wrong and it may
just be me, but I still sense a bit of bewilderment as the convergence
occurs.  Most of the time, these drafts gets discussed around a pending
meeting.  I think it is necessary to crisply define and finish this
work as soon as possible.

[1] https://www.opendaylight.org/file/odl-beryllium-diagram02
[2] https://datatracker.ietf.org/doc/draft-wang-alto-ecs-flows/
[3] https://datatracker.ietf.org/doc/draft-gao-alto-routing-state-abstraction/
[4] https://tools.ietf.org/html/draft-yang-alto-topology-06
[5] https://tools.ietf.org/html/draft-scharf-alto-topology-00

Thanks,

- vijay
--
Vijay K. Gurbani, Bell Laboratories, Nokia Networks
1960 Lucent Lane, Rm. 9C-533, Naperville, Illinois 60563 (USA)
Email: [email protected] / [email protected]
Web: http://ect.bell-labs.com/who/vkg/  | Calendar: http://goo.gl/x3Ogq

_______________________________________________
alto mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/alto

Reply via email to