Screaming Penguin - Source Does Matter

Analog is a web reporting statistics tool similar to other popular commercial and non commercial titles such as Webtrends and Webalizer.

Analog is far better than the others, IMHO, and that certainly INCLUDES the commercial ones.

Analog is blazingly fast and can report on whatever logformat you tell it to. This means it can report on ftp logs, web logs and whatever logs even in customized formats, you name it. Analog can also report on gzipped logs, which makes life much more enjoyable (web logs can get BIIIIGGGG and having a process to gzip them and or archive them makes everything much nicer, having a statistical reporting tool that can analyze them in their archived format is beautiful). Analog runs from config files that determine what to report on and how to format it, etc (and can also use command line options).

Analog is mostly numbers, not as many fancy graphs and crapola as other players, but it does have some (the new revs of Analog contain pie charts and graphs configured in the still single page view). When analog is used IN CONJUNCTION with a few of its add on partners like Report Magic, then it gets really cool. Report Magic uses conf files, stylesheets and gd to create as fancy as you want em reports from analog data that INCLUDE GRAPHS and CHARTS, etc. Using Analog to parse your logs and create the report data in a machine readable format and then using Report Magic to pretty it up makes a great combination.

I have worked on LARGE projects where Webtrends was used at several companies. Hated it. Webtrends is a hog and a half. It uses its own proprietary bloated "database" format to store data and it is very CPU intensive. In several instances in order to have Webtrends maintain statistical data for several hundred websites (virtual hosts, etc) I was forced to implement a separate machine and storage array JUST FOR WEBTRENDS. For example in order to keep a years worth of data for a site (in the verbose format the client wanted it) required around 200MB, multiply that times 300 sites and yep, you need 60GB of storage just for the silly arse reporting! Not only that but the "enterprise" rev of Webtrends really doesnt deal with more than 50 sites well at all (as explained to me by Webtrends support) and costs MEGA bucks!

Thankfully there are open source alternatives that are flat out great such as Analog.

I recently completed a project where I am using Analog and Report Magic on about a hundred sites (virtual hosts) and it works MUCH MUCH MUCH better and MORE EFFICIENTLY than Webtrends dreamed of. For the record I also tried Webalizer, but had a few issues (not able to parse out hosts from one large logfile, technically I am not using virtual hosts in the apache sense but a java web app that runs the pseudo virtual hosts for multiple websites and stores all data in one logfile by design).

Using a series of shell scripts I have a process that reads a database table and gets the site information for logging (site urls, site name, etc) and creates configuration files for Analog and Report Magic to use. Then I have a cron job that runs both Analog and Report Magic using the previously setup configuration files. Analog uses the one big logfile for all sites and parses out the data for each specific site (using a hostname element that is in the custom logformat, sites do have multiple hostnames, this is no problem). The entire process is very fast, simple and elegant. Monthly scripts are also used to archive the web logs into gzipped format and that is in the reporting config files as well so that is no problem, disk space gets saved and reports keep right on rolling.

If you need web server reporting you should definitely check out Analog and Report Magic. Here is an example of the output using the Screaming Penguin logs (note I have NOT gotten very fancy with Report Magic, this is an ultra simple version).

See the linked Analog and Report Magic sites to get started.   http://www.analog.cx/