Hi Piotr,

Thanks for starting a discussion on this scheme. Please see below.

On Sunday, March 22, 2015, Piotr Wydrych <[email protected]
<javascript:_e(%7B%7D,'cvml','[email protected]');>> wrote:

> Hi All,
>
> As some of you may remember, few years ago, long time before the protocol
> has been standardized, an idea of inter-ALTO communication has been
> proposed. As the protocol is now ready and extensions are beeing developed,
> I'd like to raise your attention to this communication scheme again :-)
>
> Since it's not possible to discuss the proposal within 5', I'll raise some
> issues on the list.
>
> 1. The rationale
> Some (maybe out-of-date) rationale can be found in
> draft-dulinski-alto-inter-problem-statement-01 from July 2011. For me,
> the two most important use-cases in which operators significantly may
> benefit from map exchange are (a) route assymmetry and (b) remote isp's
> preference.
> Regarding (a), the core ALTO protocol focuses on comparing traffic
> destination endpoints (thanks to, e.g., data from BGP tables). If an ALTO
> server is aware of routes towards the AS it belongs to, it may compare
> traffic sources between each other. In (b), I assume that information
> from remote ASes may make endpoints from the same AS (e.g., assigned to the
> same PID due the same AS-PATH) distinguishable.
>


> Since YMMV, I'd be glad to hear your voice to come out with a set of
> really important issues to be solved by inter-ALTO. I don't want to start
> an academic discussion and think of a lot of generic and specific use-cases.
>
>
Let me add one more use case, in the setting of inter-alto
communications. A setting that we are looking into is that a network is
partitioned (sharded), and hence there are multiple servers with each being
responsible for a given region. This can be considered a setting of
hierarchical alto. Hence, there are can be iterative and/or recursive
patterns on answering a query. I feel that this is an important issue, to
answer your question.


> 2. General requirements
> Base on the use-cases, a question may be raised: is there any set of
> information that has to be provided every time an ALTO server says it
> implements inter-ALTO? Some other requirements that have to be investigated
> include authentication, information confidentiality, incremental updatest
> etc.


Right. The requirements will depend on the use cases.

>
> 3. Does any IETF RFC/draft address all the reqs?
> I do not know of any.


Not I know of.

>
> 4. Extension spec
> E.g., should it be contained within one draft or split into problem
> statment / reqs / definition?
>
>
My personal preference is that a problem statement, which includes some
clear, real possible use cases can be a good starting point.

Talk to you at Dallas.

Richard


> Best regards,
> Piotr
> --
> Piotr 'GhosT' Wydrych .. xmpp:wydrych//agh.edu.pl ..
> https://urldefense.proofpoint.com/v2/url?u=http-3A__wydrych.
> net_&d=AwICAg&c=-dg2m7zWuuDZ0MUcV7Sdqw&r=4G36iiEVb2m_v-
> 0RnP2gx9KZJjYQgfvrOCE3789JGIA&m=4LLxlHCMd4UqEyyOvanuw4GrAfAT6f
> LGDAyAygWwHeo&s=aRMuAaXLNQ7_EYrl_EG_SN9EpqKEyptvEJEyhS_qdMo&e=
> A: Because it messes up the order in which people normally read text.
> Q: Why is top-posting such a bad thing?
>
> _______________________________________________
> alto mailing list
> [email protected]
> https://urldefense.proofpoint.com/v2/url?u=https-3A__www.
> ietf.org_mailman_listinfo_alto&d=AwICAg&c=-dg2m7zWuuDZ0MUcV7Sdqw&r=
> 4G36iiEVb2m_v-0RnP2gx9KZJjYQgfvrOCE3789JGIA&m=
> 4LLxlHCMd4UqEyyOvanuw4GrAfAT6fLGDAyAygWwHeo&s=
> mEoC8TS77l8YfKS4Somakf1ikOrKv40Jbfqp7FrWCA8&e=



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

Reply via email to