dcsimg

HTML page title retrieval degrades web server performance

Products

Webtrends Analytics 9.x
Webtrends Analytics 8.x

Issue

A high volume of requests from the Webtrends server overwhelms or degrades performance of the web server which generates the log files Webtrends analyzes.

Cause

This results from HTML page title retrieval being enabled either globally, or on a per-profile basis, within the Webtrends configuration. When this feature is enabled, Webtrends will index the web site at the time of analysis and pull the page titles associated with each entry in the log file(s). This feature is enabled by default.

Resolution

There are three options available for resolving this issue.

1. Reduce the number of threads used for HTML page title retrieval.

By default, Webtrends uses 100 threads for page title retrieval. Reducing this value to something much lower, such as 20, can significantly reduce the load on the web server while still retrieving the page titles to display in reports. To change this setting (and this affects all profiles globally), navigate to Administration > System Management > Hosts and select Standard Analysis Engine. Change “Use global settings” to “Use local settings,” then change “Threads for HTML Page Title Look-ups” from 100 to a lower value. The change will be immediate and the effects will be noticeable on the next analysis.

2. Disable HTML page title retrieval.

Disabling the feature globally will effectively resolve this issue. However, doing so will omit the page titles from reports. This option may be best for sites where the URLs are descriptive enough to identify the page and the burden on the web server is so great as to warrant disabling it altogether. This feature can be disabled globally by navigating to Web Analysis > Options > Analysis > General and uncheck the box for “Retrieve HTML Page titles.” To disable it on a per-profile level, edit the profile, and under Analysis > General, uncheck the the box for “Retrieve HTML Page titles.”

3. Add the WT.ti parameter to pages using meta tags.

Page titles can be appended to the URLs as parameters using Webtrends WT.ti parameter, and when log files containing these parameters are analyzed, the page titles will populate the reports above the URLs from which they originated. When page titles are added this way, HTML page title retrieval can be disabled. This method has the benefit of eliminating requests sent to the web server, but requires developers to add the title meta tags for each page that would have been indexed had the feature not been disabled. An example of how the meta tag should appear in the code, how it would appear in the log file, and how it would appear in a report is below:

In-line code:

<META NAME=”WT.ti” CONTENT=”Page Title”>

Appears as follows in cs-uri-query field in the log file:

WT.ti=Page%20Title

Displays in the Pages report (among others) as follows (assuming URL rebuilding is disabled so parameters are shown):

Page Title
http://domain.com/page.html?WT.ti=Page%20Title

Note: When using Tagbuilder to generate tags, Tagbuilder automatically includes a WT.ti entry of “Untitled document.” For each occurrence of this, ensure that the page title is changed to reflect the desired page title which would be included in the HTML title tag.

Note: If HTML title retrieval is disabled in the profile, and WT.ti tags are present on the pages of the web site, page titles will still be displayed in reports.