By Rachel Chalmers
Inktomi Corp has paid an estimated $106m to acquire Needham, Massachusetts-based WebSpective Software. Inktomi promises that the acquisition, its first in the network products arena, will strengthen its Traffic Server technology as a platform for value- added services. Dick Pierce, VP of marketing for Foster City, California-based Inktomi explains that WebSpective’s software: facilitates the intelligence for distributing content. If I’m a content provider I don’t want to have to muck around with worrying ‘Is my content in a place where people can get to it?’ I want to outsource that responsibility, preferably to someone who’s already hosting my content. Herein lies the complement to what we’re already doing. Traffic Server is the storage point that information gets pushed to.
In the classical caching model, Pierce says, the storage is there for the benefit of users pulling information down to their desktops. What WebSpective does is fill caches on the basis of network providers acting as agents acting on behalf of content providers, he says. It’s push without the drag on the desktop, as content is pushed from server to cache to await the user’s convenience. This is one element of reducing the world wide wait, Pierce argues. First we attacked the internet architecture of ‘store once’. What we’re attacking now is quality of service and guaranteed content provision.
The deal exemplifies the heat and light that has emerged from the content delivery sector since a group of MIT mathematicians launched Akamai Technologies Inc back in January 1999. Akamai and rival Sandpiper Inc run virtual networks – they have no fiber of their own, but co-locate content servers with the big network providers. But their services – FreeFlow and Footprint – have proved astoundingly successful at speeding the web to end-users. Competitors like Digital Island, Cable & Wireless and Exodus have faced the prospect of these young upstarts eating their lunch without having to make the same capital investments in infrastructure that they have already made. Little wonder that Exodus launched its own ReadyCache content delivery service earlier this week.
Pierce points out that the network providers are actual or potential customers for Inktomi’s Traffic Server. They do their secret sauce on top, but for the cache, it makes sense to use what we’ve already built, he explains. To Peter Christy of the Internet Research Group, formerly Collaborative Research, that’s just a little glib. Christy, who has emerged as an expert on content delivery, says: Let’s suspend disbelief and assume that what Akamai’s saying is true.
Akamai has placed 1000 or so servers in 40 locations on 25 networks. Their argument is that this element of the design is fundamental to their success, Christy explains. They say there’s no way to solve most performance problems on the internet, so the best thing to do is get the content on the same side of the peering point as the consumer who’s asking for it. Then it’s up to consumer’s network provider to get data from the cache to the end-user. If UUNet, say, can’t get traffic from a host server on its own network to one of its own subscribers, no one can help them. [Akamai’s model] appears to be pretty good design based on performance.
But Christy says the really provocative thing about Akamai is precisely that it is a content delivery network with no network. It calls into question whether people building network infrastructure are investing in the right place, he explains. That doesn’t just implicate Exodus, which has already overcome its crisis of faith. It also implicates Inktomi, the industry’s self-appointed Switzerland, provider of network infrastructure to all, competitor to none.
Inktomi is, in my view, a smart, aggressive company, Christy acknowledges. They aren’t going to take this lying down. But there’s only a limited amount Inktomi can do as long as its customers are networks. Those customers will be delighted by this a
cquisition, he observes, but he wonders how long Inktomi will be happy looking after the plumbing while MIT’s lucky mathematicians are drinking champagne at the industry’s most glamorous parties. In the end, he says, I might be tempted to say ‘I don’t want these customers.’ It’s all very well being Switzerland, but when Germany starts taking over Europe, what do you do? รก
