The Anatomy of a CDN

This article is very interesting, with two analysts comparing Akamai (AKAM) and Limelight (LLNW)'s relative merits and the strengths of the marketplace.

There's a lot I agree with on one side or another. So, let me summarise. Limelight provides a more flexible model that allows for the licensing of its technology in house and also provides superior support for P2P. Akamai has a more extensive, robust infrastructure (I have heard rumours of issues with the scalability of Limelight's network).

I personally think that the barrier to entry in this market are huge in terms of capital investment, strategic thinking, dealmaking and personnel - I've built a CDN and I know how difficult it is to do. I don't believe that any but the largest users of bandwidth such as Microsoft and Google will do a cost/benefit analysis and decide to in-source this function (or if they do, they're mad).

However, the key to this market is in adding value (as in any other server-client environment) by providing tools that make the infrastructure work seamlessly. Akamai has made a number of acquisitions of added value services recently, and Limelight will be forced down this path also.

There is one further consideration which is not mentioned directly in the piece and this is leverage. I worked in the media industry as a handful 0f very large media buyers cornered the market because of the volumes they could generate for media owners. Similarly, the relationship between the wholesale market, the ISP market and the reseller (ie CDN) market is set to become an interesting one with cost of scale savings being passed on to the end customer. I'm going to be blogging a lot more on this in the near future...

My experience recently in the marketplace has been that Limelight is prepared to discount far more than Akamai, but Akamai has just announced (somewhat opportunistically) support for HD off its network.