The recently released FineGround Condenser was also extensively examined during testing. We decided to cover it separately because it takes a radically different approach to content acceleration than the other products discussed. While it provides a caching facility similar to the reviewed products, it also uses an interesting intrapage caching scheme that relies on HTTP 1.1 Chunked Transfer Encoding to deliver only the changed portions of pages to the user.

The basic idea is that Condenser sets a base document and then calculates the differences (or deltas) between the base and new document, sending only the delta chunks to the user to assemble the slightly changed page. In theory, such an approach should reduce bandwidth usage much more than any of the caching solutions we reviewed.

The way such dramatic improvements are achieved is somewhat involved. First, FineGround looks at its user agent header to see if FineGround can operate. It then cookies the user if the basic requirements of condensing are met, after which it looks at the requested page, caches a base version and sends back the result. On subsequent requests of that page or similar pages, it calculates the difference and sends a version of the page that includes JavaScript and that reassembles the page by combining locally cached information with the partial data delivered to the user. Of course, this scheme relies on a browser that understands cookies, HTTP 1.1 and JavaScript. If any of these things are off, pages are not condensed, and the browser receives normal uncondensed content.

One downside of this product is that to guarantee proper condensation, accurate browser detection is required. In this area the product isn't well-developed. The tested server didn't recognize common Netscape versions, and FineGround relied on us to add the information to the configuration files to make it work with these browsers. We also found getting the product to work was just plain difficult. Installation and configuration appeared easy, but for some reason, for the longest time it just never worked. After a great deal of assistance from FineGround we found out about some features of the product we needed to get it going, most of which were just not documented.

Once we got the product running we were very impressed with the bandwidth savings and speed improvements. They were huge. FineGround makes claims of a bandwidth reduction of 40 times and an increase in download performance of up to 20 times. We found that the size of our test files were greatly reduced, but the results varied. For example, improvement ranged from just below six times (40.2K before to 7.12K byte after) all the way up to 35 times (35.4K down to 0.98K byte). Obviously, your mileage will vary, since it very much depends on the type of content being delivered.

We also ran the product as a proxy server and investigated how well it handled sites on the Internet at large. We did find that some sites just didn't seem to accelerate; however, it did not break these sites. In this sense we see the most significant aspect of this product in action, in that its gains can be had transparently. No site modifications are required, unlike many of the products we reviewed. For that alone, the $50,000 price tag may just seem worth it, especially considering the initial gains that we saw.

Unfortunately, Condenser and its approach are still very new. It's worth a careful look, but do so with significant browser testing. The reliance on JavaScript in particular, given that technology's troublesome versioning across browsers, would be enough to give us trepidation on wholeheartedly endorsing it. However, on our first look, the promise of FineGround looks very appealing.

Originally published on Network World, Published: June 11, 2001.