I was annoyed a few years ago when my hosting provider decided to remove access to the server logs. Previously, they’d provided nice pie charts as well as raw access data indicating which parts of a site were more popular. Importantly, because it was server access data, it didn’t matter whether the requested items were html pages, jpegs or other downloadable items.
The removal of this feature seems to have happened at quite a few hosting companies, otherwise how to explain the demand for Google Analytics et al.? Leave the hard work for the user’s browser, they said, and send the data to a separate, dedicated server. Server speeds have increased during the same time, so why were features taken away from users? And why, indeed, hand data over to Google et al., when you could get it from your server host?
Of course, we all realise that this is why we’re now all crying for faster multi-tabbed browsing, a demand that may have helped Google to elbow its way into the browser war theatre with its new offering, Chrome.
It’s quite possible though that the penny-pinching users themselves were at fault, unwilling to pay extra for the access to their server logs – driving down the price of web hosting by seeking out cheaper and cheaper hosts, in utter disregard of the reliability of service, and features only required by people who actually care about who looks at their web pages.
While I don’t have any data available to me right now to determine the exact root cause of the current situation, I do want to point out the duplicated effort: Web browsers do send their user agent string, identifying the exact browser version and operating system, with every http request, and in the default server configuration, access logs will usually be kept by the server, of at least the user’s IP, and the file path accessed on the server.
How many of you really feel that you need more data? Do you need to know your user’s screen resolution? Do you actually compute heat maps for your pages? Have you ever used analysis software for raw server logs, and what is wrong with your hosting company providing web access to analysis tools that are solely based on server logs?
Looking forward to your comments.