Vendor extols virtues of HTTP compression

While companies such as Packeteer, Expand Networks, Redline Networks and others have long touted the positive effects of HTTP compression, a recent study by Port80 Software (another compression vendor) shows the majority of Fortune 1000 Web sites don’t use the technology.

After conducting a survey of Web sites last month, Port80 reports that only 29 of the top 1,000 corporations’ sites use HTTP compression. Port80 says using HTTP compression could help the more than 97% of sites not using compression technology save an average of 25% on bandwidth bills and over 30 million bytes of data on their home pages. One of the top Web sites that Port80 says could benefit from HTTP compression is Intel’s.

According to the report, Intel could save 82% in costs and eliminate the need to send more than 38,000 bytes of data across its network lines if it took advantage of compression. Other sites not taking advantage of compression include retailer Dillard’s, but more interesting are companies such as Hilton Hotels and Progressive auto insurance.

Port80 reports that Hilton Hotels could avoid sending more than 100,000 bytes and reduce costs by about 54% using HTTP compression, while Progressive would save 44% and lighten its data load by more than 42,000 bytes.

Compression is a means to make data and content move more quickly over the Web. Because so much content today is dynamic, caching technologies don’t always address every company’s need for compression. There are two ways to compress data coming from a Web server: dynamically, and pre-compressed. The goal is to send less data, and faster - so compression must happen in real time and without end-user interaction.

Since smaller amounts of data (fewer packets) are being sent, they consume less bandwidth and arrive faster. Both types of compression utilize HTTP compression and compress HTML files fully three times smaller. HTML is used in most Web pages and consists of just ASCII text, which is highly compressible. Compressing HTML can have a significant impact on the performance of HTTP, research shows.