Content Provider input

["Phil Rzewski" <philr@inktomi.com>]

I wanted to kick off a discussion of a slightly less technical nature, but 
which I think is relevant given the current "requirements gathering" 
climate. It's essentially a question of scope: How much of CDNP's work 
should directly reflect the desires of Content Providers?

If we consider the "value chain" for Content Delivery, it's undeniable that 
money enters the system at both ends. Content Providers pay for hosting and 
delivery; end users pay for Internet access and paid-use rich content. The 
majority of vocal CDNP list participants are either service providers or 
technology enablers, both of whom make their living by eventually taking 
their share of these funds. Therefore, in a sense, we should all be guided 
by requirements from these parties sooner or later.

So here's a question: Are there any Content Providers subscribed to CDNP 
right now? If so, are you reading this list because you have a specific 
interest in Content Peering and how it will affect your business? For over 
a year, I've seen press and analysts refer to Content Peering as a concept 
without giving any indication they knew what it was. I've even seen 
high-level spokespeople for Content Providers (e.g. news authorities) say 
that Content Peering is important to them. However, as you can imagine, 
technical requirements were strangely absent from these articles. :)

I'll certainly kick things off by offering my own data point. Some of you 
may have seen the drafts I submitted before the last IETF, namely 
"Origin/Access Content Peering for HTTP" and the two companion drafts. 
While these were relatively simple, they did evolve out of some Content 
Provider requirements. They proposed a reasonably simple model of 
Distribution and Accounting. The Request-Routing was a "null case" because 
access providers' proxy deployments became the initial target set of 
"surrogates". Content Providers would maintain a relationship with a single 
Hoster or CDN that would, in turn, fan out Content Signals for Distribution 
and aggregate Accounting data in return.

By comparison, I might see the proposals to coming out of CDNP as a "next 
generation", since they offer more possibilities. Content Providers can be 
considered as a form of "degenerate CDN" that can act as an authoritative 
Request-Routing entity and therefore control delivery from multiple CDNs. 
Some of the negotiation and metrics we've been debating could allow 
reasonably robust control. In fact, in some of my contributions, I've 
mentioned how these could allow Content Providers to construct robust 
delivery policies (use one CDN for regional, one CDN for "default transit", 
etc.) However, I'll be the first to admit that I'm speaking of those 
concepts from a "wouldn't it be cool" concept. I have yet to have a Content 
Provider come to me and say "I wish I had a Request-Routing system that 
allowed me to exercise this level of control".

So I'll close with the question again. Are there Content Providers here 
that have an opinion on this topic? If not, do we care to seek them out? Or 
is the possibility of having a solution for "power user" Content Providers 
just a coincidental side-effect of designing a solution by CDNs, and for CDNs?

Just food for thought.

--
Phil Rzewski - Senior Architect - Inktomi Corporation
650-653-2487 (office) - 650-303-3790 (cell) - 650-653-1848 (fax)