This section explains how to make analysis focused on the types of users who access the Web sites.
Analysis focused on users enables the following information to be recognized and utilized for Web site operation.
Which users often access membership sites?
Which companies often access Web sites?
Systemwalker Service Quality Coordinator identifies users as follows:
Identifying users by user name authenticated on the Web server
If authentication is performed on the Web server, authorization user names can be used as a key for analysis.
Identifying users by IDs set in a cookie
If authentication is performed on the Web server using CGI, no information about authorized users is output to the Web server log. You can perform analysis that identifies users by setting user names authenticated by CGI in a cookie.
If authentication is not performed by CGI, you can perform analysis that identifies users by setting for Cookie an identifier that identifies the user on each page of the Web service.
Identifying users by client host
When the Web server performs no authentication, such as that by the service that publishes information, and no information is set in a cookie, then users can be identified by the client host.
For the identification of users by client host, the ID address and host name of the client is used.
Note
For analysis to identify users, content must be created such that authentication is performed or a cookie is set reliably when the Web service to be analyzed is used. If a page in the Web service is accessed directly without authentication or Cookie settings, then analysis that identifies users cannot be done.
Analysis that identifies users with an ID set in a cookie cannot be done if the user blocks cookies with a browser setting. In such a case, analysis that identifies users by client host can be used.
Note that in analysis in which users are identified by client host, different users are handled as the same client host in the following cases:
IP addresses are allocated dynamically by DHCP (for example, ISP is used to connect).
Access is via a proxy server (for example, corporate users connecting from behind a firewall).
The following environment settings are required to analyze what kind of users visits the Web sites:
Setting a Web server log output
Setting a usage DB environment definition file
To analyze what kind of users visited to the Web server, the Web server must be set so that information about the users is output to the Web server log.
Identifying users by user name authenticated on the Web server
Information about the user names that are authenticated by the Web server is output to the Web server log.
If the information is not output, set the Web server so that it is output to the Web server log.
Identifying users by IDs set in a cookie
If authentication is performed on the Web server using CGI, and user names authenticated by CGI are set in a cookie, set the Web server so that the cookie information is output to the Web server log.
The following gives a setting example for the Web server.
Microsoft(R) Internet Information Services 6.0
Chose from the [Start] menu as shown below to start Internet Service Manager.
[Start]
-> [Programs]
-> [Management tool]
-> [Internet Service Manager]
Note
If the environment is not as shown above, perform operation appropriately for the environment.
Set extended log properties.
Point the mouse to the Web site for which extended log properties are to be set, then display the extended log property setting window.
Check the following and press the [Complete] button.
Cookie ( cs(Cookie) )
Save the Web server log.
If the information to be output to the log is changed in the extended log property setting window, the log output format is changed.
Note
Because the new log output format is defined for Systemwalker Service Quality Coordinator, it differs from the prechange log output format, causing an analysis error. Therefore, save and delete the prechange log.
Apache
[Windows]
Chose from the [Start] menu as shown below and open the configuration file.
[Start]
-> [Programs]
-> [Apache Web Server]
-> [Management]
-> [Edit Configuration]
Note
If the environment is not as shown above, perform operation appropriately for the environment.
Set the log output format.
Add the lines shown below to the log-related definition part.
The following gives an example when cookie information output definitions are added to the common format, a default log format.
#CustomLog logs/access.log common * Define this line as a comment.
CustomLog logs/access.log "%h %l %u %t \"%r\" %>s %b %{cookie}i"
See
Refer to the Apache manual for details.
Reflect the settings.
Save by overwriting and quit the editor. If the Apache HTTP server is active, shut down and reboot it.
[UNIX]
Use the editor to open the configuration file.
Set the log output format.
Add the lines shown below to the log-related definition part.
The following gives an example when cookie information output definitions are added to the common format, a default log format.
#CustomLog logs/access_log common * Define this line as a comment.
CustomLog logs/access_log "%h %l %u %t \"%r\" %>s %b %{cookie}i"
See
Refer to the Apache manual for details.
Reflect the settings.
Save by overwriting and quit the editor. If the Apache HTTP server is active, shut down and reboot it.
Identifying users by client host
Normally, information about the client hosts is output to the Web server log.
If the information is not output, set the Web server so that it is output to the Web server log.
Set "Format" that is defined in the analysis target log definition block in the use trend DB environment definition file, based on the user identification method.
Identifying users by user name authenticated on the Web server
Specify tokens (c-user) for Format, based on the Web server log output format.
Identifying users by IDs set in a cookie
Specify tokens (c-cookie) for Format, based on the Web server log output format. When the tokens (c-cookie) are specified, define the cookie format according to the contents that are output as cookie information.
The following gives a definition example.
Output format of Web server log cookie information
"ID001;20020401" (ID001: user name)
[Server] Symbol = PUBLIC <-- omission --> [Log] Symbol = WWW <-- omission --> Format = "s-time{yyyy-mm-dd HH:MM:SS} c-host s-method s-url s-status s-bytes \"c-cookie{id;*}\"" <-- omission --> |
Identifying users by client host
Specify tokens (c-host) for Format, based on the Web server log output format.
Specify analysis methods in use trend analysis window, based on the user identification method.
Identifying users by user name authenticated on the Web server
Specify "Authenticated user name base" as an analysis method in the use trend analysis window.
Identifying users by IDs set in a cookie
Specify "Access ID base" as an analysis method in the use trend analysis window.
Identifying users by client host
Specify "Client host name base" or "Client IP address base" as an analysis method in the use trend analysis window.
The following shows the window for analysis with users identified by client host.