Does that mean that an ECS should NOT be declared as "using" a particular
map? That works for me. That would mean the server is free to select the
"appropriate" map for that set of endpoints. Or some other mechanism
altogether.
But if so, the protocol should say that. How about replacing the second
paragraph of {10.4.1} with the following (additions delimited by **):
"It is important to note that although this resource allows an ALTO Server
to reveal costs between individual endpoints, an ALTO Server is not required
to do so. A simple alternative would be to compute the cost between two
endpoints as the cost between the PIDs corresponding to the endpoints,
**using one of the network maps defined by the server**. See Section 14.3
for additional details.
"**Although the Endpoint Cost resource may be based on a specific network
map, the resource MUST NOT be declared as depending on that network map (see
the "uses" attribute in Section 8.5.2).**"
From: "Y. Richard Yang" <[email protected]>
Date: Fri, July 26, 2013 14:30
To: Wendy Roome <[email protected]>
Cc: IETF ALTO <[email protected]>
Subject: Re: [alto] Endpoint Cost Service & Network Maps
Hi Wendy,
Another good observation. If I remember the intension, the "omission" was
intentional. Given your feedback, I think we should add a sentence to say
that ECS has internal implementation flexibility, using publicly announced
network/cost maps, internal maps, or other mechanisms.
Richard
On Jul 26, 2013 2:05 PM, "Wendy Roome" <[email protected]> wrote:
> Does it make sense to allow an Endpoint Cost Service (ECS) resource to
> "use" a Network Map resource?
>
> If it does, then that ECS is equivalent resolving the endpoints to PIDs
> with that map, and getting the costs from the associated cost map.
>
> But if an ECS resource does not "use" a network map, then that ECS is
> based on some unspecified source of cost data. It might use one of the
> network & cost maps, or it might use some other mechanism altogether.
>
> As far as I can tell, the current draft doesn't say anything about this --
> it neither forbids nor requires an ECS to depend on a network map. The
> only guideline is that the ECS example in {8.5.4} doesn't have a "uses"
> attribute. But I wasn't sure if that omission was intentional.
>
> - 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
