One of my favorite quotes (attributed to Adm Grace Hopper of COBOL fame) is:
"There is nothing new in computing, only new ways of doing old things."
While I get constant reminders of this sage observation, the latest was an article on Matt Marshall's Silicon Beat about a new company called Itiva that is launching a product to provide caching for video content. Their value proposition is to lessen the bandwidth costs for distributing content across the network to viewers. Matt's article provides an overview of how Itiva uses their cache to avoid re-transmissions.
We've seen this show before; in the late '90s as Internet usage was ramping quickly, several companies introduced caches for HTML transported content. I was an investor in one of them, Cacheflow (now Blue Coat Systems) that provided high performance web caches that were used by ISPs to reduce their bandwidth requirements and also to provide better response times to their users. My friend Jonathan Seelig co-founded Akamai as a service provider building out a network of distributed caches to help companies such as Yahoo scale and provide better service. I remarked on the rise in Akamai's business over the last year or so on a recent post on Net Neutrality.
There is definitely room for companies like Itiva. To me, its a sign that we're beginning to loop back around the Internet virtuous circle; first you build the core of the network, this enables you to build new applications that leverage the bandwidth and reach of the network. As the application grows, bandwidth costs begin to dominate the scaling issues and so the natural response is to distribute content (and functionality) to the edge of the network. The success of the application encourages new applications to be developed and at some point causes the core to be upgraded as bandwidth becomes a limit to growth.
There was dramatic investment in the core of the network in the late 90's, carrying through to the unprecedented pull back in Telecom spending in 2001 as it became apparent that the network had been enormously overbuilt. At the time several analysts projected that there was enough capacity in the core of most networks to last at least until 2007.
But as much as the cost per bit of bandwidth drops, the true cost of distributing content comes home to roost when the size of the content increases. We saw this before as the web went from mostly text based pages to graphics; now we're going from still graphics to video content (and wait until the content goes HD!) and the mass number of bits to be moved increases dramatically!
The time is perfect for another set of innovations (and investments!) at the edge of the network.