HTTP Load

It has been observed that several misusers out there like downloading things. When they find a thing they like, they download it all the time.

One of them was observed during the period of three days. He downloaded the same five files over and over again, using as much bandwidth and as many parallel connections as the server could manage. He downloaded those five files thousands of times during the observation time, files which haven't changed in between.

After three days new rules have been added to the firewall to block the misuser. That didn't impress him, he continued to send HTTP requests for another day and a half despite the fact that these were not answered any more. Since no human would click the reload button twenty times per second for more than thirty hours, it must have been an automated process.

This misuser won't download anything from this site ever again, but he isn't alone. Dosens of others have been observed doing the same thing. During just one day, any one of these can produce traffic the site wouldn't have in a month otherwise.

I pay for my traffic. And I won't pay for its senseless waste. Please understand that this isn't a high-volume portal but a small private site operated by an individual. It is hosted on a computer with a fast connection, so the misusers described above can consume my budget if I let them.

In the beginning, the speed in the HTTP download area was reduced to 8 kB/s. While this worked well in the old Apache 1.3 environment, the restriction isn't easily established with the currently used Apache 2.

Well, rough times call for rough actions so I decided to shut down the HTTP download area for good. The download area will continue to operate through FTP, where the bandwidth and the number of concurrent connections can be easily controlled.

Webmaster, 10. March 2004