1. What do developers need to know to use SpeedTracer?

2. Where can I find more information about NCSA HTTPd?

3. Hit counts are considered unreliable. How does SpeedTracer provide more reliable user statistics?

4. Does SpeedTracer also provide hit count as a comparison?

5. Does SpeedTracer rely on "cookies" or user registration to identify user sessions?

6. Can SpeedTracer identify user sessions if a proxy server replaces user identities and browser caching?

7. How does SpeedTracer define a user session?

8. What kinds of log data does SpeedTracer need to perform the Web usage?

9. Does the HTTPd configuration need to be changed to generate these log files?

10. What statistics does SpeedTracer generate?

11. What are entry and exit pages?

12. What are the differences between path reports and group reports?

13. What do the error messages "Cannot open file for debug" and "Cannot open file for storing" mean?

14. Why do the error messages like "Cannot open access_log file" and "Cannot open referrer_log file" mean?

15. What happens when the frequent external referrals look like the URLs of my own Web site?

16. What do I need to know to use SpeedTracer?

17. Why does SpeedTracer seem to take so much time to finish in the first few times when the IP host names are not available in the log?

18. What's new on the latest update of SpeedTracer?


1. What do developers need to know to use SpeedTracer?

To use SpeedTracer, developers should be familiar with NCSA HTTPd, and more specifically, HTTPd with the LogOptions Directive.

[Goto Top]

2. Where can I find more information about NCSA HTTPd?

For general information about NCSA HTTPd, go to hoohoo.ncsa.uiuc.edu.

For specific information about the NCSA HTTPd LogOptions Directive go to: hoohoo.ncsa.uiuc.edu/docs/setup/httpd/LogOptions.html.

[Goto Top]

3. Hit counts are considered unreliable. How does SpeedTracer provide more reliable user statistics?

Instead of simple hit counts, SpeedTracer uses innovative inference algorithms to identify user sessions. Advanced data mining algorithms are then applied to identify frequent traversal paths and most often visited group of pages.

SpeedTracer provides user-based Web usage statistics for most commonly visited individual pages, group of pages (maximum number of pages is five), external Web site referrals, and user traversal paths with four links.

[Goto Top]

4. Does SpeedTracer also provide hit count as a comparison?

Yes, SpeedTracer provides both hit counts and user counts on the frequently visited pages.

[Goto Top]

5. Does SpeedTracer rely on "cookies" or user registration to identify user sessions?

No, SpeedTracer does not rely on "cookies" or user registration for user session identification. SpeedTracer does not specifically identify any particular user identity. It only provides aggregate user-based statistics.

[Goto Top]

6. Can SpeedTracer identify user sessions if a proxy server replaces user identities and browser caching?

Yes, SpeedTracer uses all available information from the access, referrer and agent logs to reconstruct user traversal paths for identifying user sessions. SpeedTracer also takes into account the special cases, such as when a browser repeatedly clicks "reload," "forward," and "backward."

These identified sessions are not 100 percent accurate, but they represent very good estimates. They are much more precise than tracking unique IP addresses from the access log, which are generally proxy server IPs.

[Goto Top]

7. How does SpeedTracer define a user session?

A user session is the sequence of consecutive page references issued by a user to your Web site. However, if a user does not make a request for a pre-specified period of time, the previous references from the user are considered a completed user session.

[Goto Top]

8. What kinds of log data does SpeedTracer need to perform the Web usage?

SpeedTracer uses three log files: access, referrer and agent.

[Goto Top]

9. Does the HTTPd configuration need to be changed to generate these log files?

For NCSA server: Yes, the LogOptions Directive should be set to "Combined". For Apache server: Yes, the following line should be included in the server httpd.conf file for specifying access log file.

CustomLog logs/combined_log "%h %l %u %t \"%r\" %s %b \"%{Referer}i\" \"%{User-agent}i\""

For ICS (Lotus Domino) server: No change is needed.

[Goto Top]

10. What statistics does SpeedTracer generate?

SpeedTracer generates three types of reports: user reports, path reports and group reports.

User reports provide session-based statistics, such as frequently visited pages, entry pages, exit pages, distributions of session durations and session sizes.

Path reports show the frequent traversal paths of different lengths.

Group reports track frequently visited groups of up to 5 pages.

[Goto Top]

11. What are entry and exit pages?

An entry page is the page at which a user begins a session.

An exit page is the last page visited in a user session. Exit page statistics will indicate what information on a site is being skipped or which advertisements lead users out of your Web site.

[Goto Top]

12. What are the differences between path reports and group reports?

A user traversal path shows pages visited consecutively through their HTTP links. A group of pages visited during a user session may not all be on a single traversal path. There may be more than one traversal paths in the group, or the pages may be visited at different order through various intermediate pages.

[Goto Top]

13. What do the error messages "Cannot open file for debug" and "Cannot open file for storing" mean?

A "w" privilege on the report directory needs to be created so that SpeedTracer can create temporary working files during its execution. Check if permissions are properly set, that the report directory is correctly specified in "speedtracer.conf" file and the proper command line option is included when SpeedTracer is invoked.

[Goto Top]

14. Why do the error messages like "Cannot open access_log file" and "Cannot open referrer_log file" mean?

SpeedTracer needs the file names of access, referrer and agent logs as well as the name of the directory where they reside. This information must be specified in the "speedtracer.conf" file and be included in the command line option when SpeedTracer is invoked. Check to see if the file names and the directory are misspelled.

[Goto Top]

15. What happens when the frequent external referrals look like the URLs of my own Web site?

This may indicate that the IP address of your Web site and its full host name and aliases are not correctly specified in the "ip.file." Make sure you invoke SpeedTracer with the proper command line option.

[Goto Top]

16. What do I need to know to use SpeedTracer?

To use SpeedTracer, developers should be familiar with NCSA HTTPd, and more specifically, HTTPd with the LogOptions Directive.

[Goto Top]

17. Why does SpeedTracer seem to take so much time to finish in the first few times when the IP host names are not available in the log?

SpeedTracer contacts remote name servers to resolve the IP host names if they are not available in the ip.db file that is included in the download package. However, once resolved, the names are stored in the ip.db file in the reports directory. Execution time should improve dramatically subsequently as IP host names can be resolved locally. Please do not remove the ip.db file from the reports directory because it will greatly degrade SpeedTracer performance.

[Goto Top]

18. What's new on the latest update of SpeedTracer?

An enhanced GUI, additional reports and features are available in the latest update. There are table, pie and bar charts for each report. Additional reports include an executive summary, two distributions of user session widths and depths, and two demographic reports based on organization and country. Additional features include session-based inclusion/exclusion based on URL, method, host name and return code. One can have reports generated for all sessions that include or exclude a specific URL, method, host name, or return code, or a combination of them. Note that these updates are available for the ICS logs and for the NCSA/Apache logs in the combined log format.

[Goto Top]