On Thu, Jul 18, 2013 at 12:34 PM, Wendy Roome <[email protected]>wrote:

> Unless I missed something, draft 17 allows an ALTO Server to provide more
> than one full cost map resource for the same cost-type (and network map).
> What would that mean?  I can think of a couple of possibilities:
>

To set the stage properly, the protocol draft never prohibited two cost
maps of the same cost type with different data even before draft -17.


>
> * That should be illegal.
>
>
> * The resources are equivalent, and they return the same cost map (at
> least eventually :-)). The duplicates provide load balancing, so the
> client should pick one randomly.
>
> * The resources return different cost maps. Eg, the server has several
> different versions of one cost-metric, and they are distinguished by the
> resource id. This implies that a filtered map service that offers a
> multi-version cost-metric should have a 'uses' attribute to indicate which
> cost-map resource it's based on.
>
> I think the last interpretation will cause a lot of confusion. Invent a
> new cost-metric name instead!
>
> As for the other two, I have a mild preference for outlawing duplicates,
> but I'd support the separate-but-identical interpretation if the rest of
> you like that idea.
>

I agree with disallowing duplicates, with the caveat that extensions can
introduce distinguishing attributes in the future [will need to update the
draft to document how that is done].  For example, imagine an extension
introduced a 'granularity' attribute; it would be good to allow ALTO
clients to have routingcost at multiple granularities in the same IRD.

Rich


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

Reply via email to