Aug 22 2007
I got a call today from Damieng to warn me that my site bandwidth might get chewed up as he decided to move some image files from a popular blog post of his to a subdomain on my hosting server. It seems that his home DSL line was being saturated and he desperately needed to reduce the load. Whilst the activity was killing his DSL line it hardly made a dent in my monthly bandwidth allowance. One interesting thing I did notice whilst poking around in my log files was the user agent string that Google uses for its rss feed fetching service. When users are reading rss feeds in Google Reader the feeds are not being fetched real time. Instead Google has a feed fetching service that checks feeds every 30 minutes or so and caches them, users then subscribe to these caches. This reduces the load on the server hosting the rss feed, whether you have 1 subscriber or 1000 subscribers, the load doesn’t go up. But that can play havoc with web stats. How can you know how many people are subscribed to your rss feeds? What Google has done is make the user agent string for the feed fetcher dynamic, the dynamic part of it being the number of users that have subscribed to your feed through Google Reader. Here’s an example from my own logs:
Feedfetcher-Google; (+http://www.google.com/feedfetcher.html; 15 subscribers; feed-id=3000255605479237567) Whilst looking for the again in my logs, I also noticed that BlogLines, NewsGator and a few others also seem to be going the same thing. I wonder if there are any stats applications out there that actually make use of this information.