RE: Distribution CPG Protocol - Some Thoughts

[Oliver Spatscheck <spatsch@research.att.com>]

Stephen Thomas writes:
 > At 04:52 PM 2001-01-04 -0800, Tomlinson, Gary wrote:
 > >There's a couple of subtle issues in this scenario that need to be
 > >addressed.  There are at least three kinds of replicating surrogates that
 > >I'm aware of, full set replica, demand driven caches and pre-populated
 > >(pinned) caches.  Additionally, surrogates can be categorized by the
 > >services they offer, HTTP, streaming media (Real, MMS, QuickTime, MPEG, etc)
 > >etc. What this means is, surrogates are not just generic delivery points,
 > >but rather are likely to be more specialized.  Given this complexity along
 > >with the implementation policy differentiators in CDNs, I've been working
 > >under the assumption surrogates would be represented in aggregate by their
 > >CDN as a set of service capabilities, such as capacity, proximity, delivery
 > >service modalities, etc.  With this in mind, the advertising of content
 > >isn't directed at specific surrogates but rather to CDNs who in turn manage
 > >their surrogates to meet the commitments the entire CDN signs up for.
 > 
 > Gary has said exactly what I was trying to, but much better. I, too, was 
 > thinking that a content provider signs up with a CDN (Or one CDN signs up 
 > with another CDN), and it was up to the CDN to manage/operate it's 
 > surrogates. I think the idea was that the CPG would actually be the one 
 > participating in the protocol, not the surrogates directly. In which case 
 > the CPG would probably represent an aggregate capacity and set of capabilities.
 
As of Gary's other comments and my follow up it seems still open if
the content gateway or accounting gateway will provide this information.
I would like to hear from more people how they feel about this.
I think this decision is very important in our design in that
we don't need a BGP like interface for the CPG if all information
is gathered by the accounting system. In this case we know which CDNs
will be used a priory and all we have to do is to inform them 
where to fetch content. We don't have to advertise the content to all
CDNs (or surrogates as originally proposed).

Anyway I think a total aggregate is not good enough. We have to be able
to identify regions (or ASs or whatever else we call it). I think the
case in which two CDNs are in the same region (from a network perspective)
and the redirection system has to decide which one to use is an
important one we have to address. It also is a open question if we
should inform a CDN in the CPG that clients in some regions will
not use its service or let the CDN figure that out by itself.

Overall it is a fine balance between allowing a CDN to manage/operate
its surrogates by itself and to gather enough detailed information
on the gateways to make any intelligent decision. I don't
think a single number for all surrogates per CDN will cut it. I also
agree that information on a per surrogate basis while desirable
is probably unrealistic. However, aggregated information for
meaningful network regions seems like a reasonable approach to me.


Oliver