Ca$hing In

For my money, the hottest market segment of 1999 wasn't storage, although it was storage-related. That segment was Internet caching. If only I had invested in Akamai Technologies, the Israeli Web caching company with MIT roots. A little money down on the day the company held its initial public offering (IPO), and I would be writing from a beach somewhere. Akamai has come down to earth somewhat, but the company has made a lot of people's lives "comfortable." Somehow, a new company with good but not breakthrough technology instantly achieved a market cap bigger than a company (Novell) with an entire portfolio of networking products, one of which is an Internet caching product the equal of Akamai's that takes my breath away. Long ago, the stock market stopped making sense to me.

Something in the air must have convinced several companies I've contacted lately that they can make significant sums of money selling cache compression solutions. Or is it that Akamai and other leaders in this new technology of caching on the Internet haven't done everything to optimize the technology?

Data transferred from a Web server to cache servers and then onto yours isn't significantly compressed. And because the Internet consumes large quantities of storage (with phenomenal storage growth rates), one could reasonably argue that cache compression will save companies a lot of money. Compression also makes transfers faster.

But I thought companies could purchase compression technology from vendors "out of the box" or easily modify a standard routine. Despite new wrinkles in writing compression algorithms, people adapt most of them from a few general classes of algorithms. Every so often, some math wiz comes along and figures out a new algorithm that changes things, but that doesn't happen often (and is a big deal when it does). Compression is a core enabling technology that makes new things possible. So fractal compression, new JPEG and Moving Picture Experts Group (MPEG) standards, and the stuff C-Cubed did a few years back are important.

However, Internet cache compression involves modifying standard algorithms to "understand" the multiple data types with which browsers work. A couple of vendors have indicated that cache compression is a considerable art.

I saw one of these products at Citrix's iForum when I spoke with Packeteer, a 4 1/2-year-old Cupertino, California, company. Packeteer first caught my eye 3 years ago with a bandwidth monitoring and control product called PacketShaper that is perfect for the thin-client/server market. In July, Packeteer acquired the British Columbian company Workfire Technologies, which has a cache compression product.

The result is AppCelera Internet Content Accelerator ICX-55. The art, according to Jeff Barker, Packeteer's director of Product Marketing, involves a proxy caching engine that uses a content-aware compression algorithm to produce dynamic variant caching. (The product is aware of content, browser, and speed.) The $10,000 hardware device accelerates Internet application performance by 35 percent to 50 percent and supports up to 200 connections per second. This solution requires no changes in either clients or servers. AppCelera is in very late beta and should be available at the end of November. Other companies in this space include Redline Network and Boostwork. So expect to see more products in the coming months.

My woeful saga of Internet connectivity continues. My cable modem arrived at home and works every so often for short periods of time. After AT&T Broadband's (nee Media One) fifth repair visit, we got a higher-capacity wire from the street to the house, the last hundred yards, but the system still doesn't work. On the office front, the third ISP is attempting to install DSL. Theoretically, the phone number works. I'm going with DSL provider number three for the sheer sport of seeing whether anyone can give me DSL. No wonder 75 percent of us are still on dial-up and cache compression is so useful.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.