HTTP-ANALYZE Product Description

HTTP-ANALYZE PRODUCT DESCRIPTION

What is
http-analyze?

http-analyze is a logfile analyzer for web servers. It analyzes logfiles and creates a comprehensive statistics report in HTML and VRML (3D) format. The report contains summaries of all accesses sorted by various criterias in tabular and in graphical form. http-analyze has been optimized to process large logfiles as fast as possible. The analyzer recognizes HTTP/1.1 server responses as well as the CORE top-level-domains. Any other domain can be added in a configuration file. The software comes in pre-built binary form for easy installation and is also available in source form for compilation on any platform which supports an ANSI-C compiler.

How do I use
http-analyze?

Requirements

http-analyze can be used with any web server supporting the NCSA Common Logfile Format (CLF) or the W3C Extended Logfile Format (ELF) such as Apache, iPlanet and almost any other modern web server - except Microsoft's IIS, which uses a proprietary logfile format. For IIS we provide a perl script to convert logfiles in IIS (proprietary) format into standardized W3C ELF format. See the manpage of http-analyze and your server's documentation for more information about customizing the server's logfile format.

http-analyze is best utilized by running on the same machine the web server is running on. This means that the webmaster/administrator of the web service (not the end-user!) should install http-analyze and configure it to be executed automatically on a regular base, for example once or twice per day. Although you could use http-analyze on your own PC to process a raw logfile on your local hard disk, this is not the recommended usage, since you would have to transfer the logfiles to your system first and then to execute http-analyze manually to create a statistics report. However, this (unsupported) usage by end-users is generally possible.

Installing the executable

To install the software, download the pre-built binary for your platform and install it using the software installer or the installation script provided in the distribution. If a pre-built binary for your platform isn't available or if you want to generate a specially tailored version of the analyzer for your platform, you can download the source distribution of the software and compile it for yourself or have us compile a version for your platform (commercial licensees only).

Special note for Windows users

For Windows systems, the software comes as a self-extracting ZIP archive with a proprietary installation and setup utility. Except for the installation, the software is to be used as a console application. There is no graphical interface to just execute this program - the GUI is the statistics report created by running http-analyze automatically on a regular base! In other words: On Windows the program is used in exactly the same way as it is used on Unix systems, so the same documentation for Unix applies also to the Windows version. To use http-analyze on Windows, make note of the installation path, open a MS-DOS Command Prompt, change into the installation directory and run the executable file http-analyze.exe by typing its pathname. To install it completely, set up an AT job, which runs a batch script to executes http-analyze on a regular base. Since the set-up of such a scheme needs to be done just once and is usually never changed lateron, there is no GUI to run the http-analyze executable, since it is just not needed.

Testing the analyzer

For a first test, you can run the analyzer out of the directory where you unpacked the files using your web server's logfile. Next, you will probably configure your operating system to have it execute http-analyze on a regular base to create an up-to-date statistics report automatically for your web servers. On Unix systems this is done using the cron service and on Windows systems using the AT scheduler. Please consult the documentation for your operating system for more information how to have programs executed in regular intervals.

A note on logfile rotation

Included in the distribution there are two shell scripts: one for rotating logfiles and one for running http-analyze for a set of pre-defined web servers. For a full statistics report it is important that you keep all logfile data available for a whole month before rotating (moving away) the logfile. This means that logfiles should be rotated at the 1st of the next month. Next, after rotation this logfile has to be »finalized« by creating a final statistics report (the first report which will not change anymore). Therefore, the rotate script executes the script to run http-analyze with the old logfile and thereafter with the newly created logfile. On all other days except the 1st, just have cron execute the script to run http-analyze for all web servers. If you prefer own scripts for rotation of logfiles and execution of http-analyze, strictly use the scheme described above, otherwise you're statistics reports will become zeroised each new month.

Configuring the analyzer

To change certain default settings of http-analyze, create a config file using http-analyze and customize the settings in this file. If you already have a configuration file from an older version of http-analyze, it can be converted automatically into a new one. See the section "Configuration File" in the documentation of http-analyze for more information about how to create or convert configuration files.

Please read the options in the manual carefully. http-analyze uses only a very small subset of defaults suitable for most simple web server installation. For example, to have http-analyze recognize PHP or ASP files as documents (or more correctly: as pageviews), you would have to define the PHP/ASP suffixes in the configuration file explicitely. Most confusion about http-analyze not obeying certain file types results from a missing configuration.

Viewing the statistics reports

For historical reasons and to retain compatibility with previous versions, there are three interfaces to the statistics reports:

the conventional interface: can be viewed by using any browser which supports tables. If JavaScript is enabled, a small, separate navigation window will pop up. If JavaScript is disabled, the navigation links will appear in the same window as the statistics pages.

the frames-based interface: can be viewed with any frames-capable browser which also supports JavaScript. May or may not work with certain versions of MSIE due to Bill's will.

the 3D (VRML) interface: requires a VRML 2.0 browser such as the CosmoPlayer plug-in (tested with Netscape Navigator 3, Communicator 4, MSIE 4).

Browsing the 3D scene

Cosmo Software To view the 3D models which http-analyze generates, you need a VRML-compliant browser plug-in such as CosmoPlayer. http-analyze generates VRML 2.0-compliant code, which is known to work with CosmoPlayer 1.1 and 2.1. Note: You do not need the CosmoPlayer plug-in to view the statistics report in its normal tabular and graphical (2D) form.

Update January 2006: CosmoPlayer was available for Windows and MacOS, but unfortunately has been discontinued by its latest owner Computer Associates. SGI users still can download CosmoPlayer 2.1.5 for IRIX from SGI's Download Cool Evaluation Software web page.

Registration of http-analyze

The software is free for personal and educational use. A Commercial Service License is available for professional use with support and updates as often needed by ISPs and companies. If you plan to make money with http-analyze - either directly or indirectly by adding value to your services - you are required to buy a Commercial Service License (once). We can continue developement and support for http-analyze only if people support us by registering the software if used in a commercial installation. Please see this page for more information about the availability and prices of different license types.