Statistics for GWebCacheThese are some simple statistics for the GWebCache script and the amount of traffic it generates, because people were interested to see how much traffic it will generate for their site. Updated! 5/13/2002 - I finally got around to using a traffic sniffer to monitor exactly how much traffic is generated by the cache. Also I am now making my cache scan reports available! Traffic Statistics: Using a traffic sniffer, I was able to monitor the amount of traffic generated by the script. I sent the script around 3000 hostfile, urlfile, and update requests (i.e. 1000 of each - because these are the only requests that are sent by clients). On average, each request took 3.7kb of total traffic. With the same test I was also testing the average execution time - on the 200MHz computer (Linux+Apache) I was using as a server, the script took an average of 0.2 seconds to execute. (Note: Neither the CPU or the network (10Mbit) was maxed out. Increasing the requests sent per minutes does not change this statistic - remember that file locking is used, and all requests will therefore be processed more or less in serial. If the load gets too high, clients will simply time out.) Calculations: Currently (5/13/2002) there are 71 web caches out there, 36 of which are functional. An average (functional) cache receives about 17 update requests an hour. The number of hostfile and urlfile requests is on average the same as the number of update requests, therefore an average script receives 51 requests per hour. At 3.7kb per request, that is about 200kb/hour, 4.7MB/day, or 140.5MB/month. (Yes, I rounded a lot, but only up.) Conclusions: Most websites give traffic allowances in the gigabyte range - so as you can see, the traffic generated by running a cache is minimal. Processor usage is absolutely minimal. Predictions: The more (functional!) web caches there are, the less traffic per web cache. However, we might soon see Morpheus taking the new Gnucleus code, which will mean a sharp increase of traffic. The only way to reduce the traffic in such a case is to reduce the frequency at which the clients send updates to the script, or publicize the web cache so well that we increase the number of caches out there by a lot :-)
GWebCache Home |