Re: [earlywarning] What problem is ATOCA trying to address?

so right off the top of the head, one could envision the following distribution models:

a) end-to-end unicast -- worst case scenario in terms of load concentration at a single source
b) overlay unicast segments -- retaining the the unicast only model, but pushing the distribution to a collection of nodes operating as an overlay.  Headaches include management of the overlay, but potentially higher resiliency
c) unicast/multicast servers -- can be considered an extension of (b), but in this case, include a set of nodes that take the unicast traffic and forwards it downstream via multicast to the end nodes.  There is a fair amount of work that has been done with this model, and it leverages the efficiency of multicast, where multicast is likely to only reside in access/enterprise networks

-ken

On Mar 26, 2010, at 12:26 PM, Art Botterell wrote:

> On Mar 26, 2010, at 3/26/10 8:44 AM, Brian Rosen wrote:
>> We're not at the mechanism state yet.
> 
> True, and that's not really what I meant.  My point was that the shapes of the problem and the constraints seem similar to the more general problem of multicasting, on which there's been a lot of prior work.
> 
> So perhaps it would be useful to enumerate the particular ways in which this problem ISN'T just an instance of multicasting.
> 
> - Art
> _______________________________________________
> earlywarning mailing list
> earlywarning@ietf.org
> https://www.ietf.org/mailman/listinfo/earlywarning