Analysis of data

After you collect data from multiple sources, you must determine how the data can help you to resolve your particular problem.

To analyze the data, take the following actions:

• Determine which data sources are most likely to contain information about the problem, and start your analysis there. For example, if the problem is related to installation, start your analysis with the installation log files (if any), rather than starting with the general product or operating system log files.
• Have a clear understanding of how the various pieces of data relate to each other. For example, if the data spans more than one system, keep your data well organized so that you know which pieces of data come from which sources.
• Confirm that each piece of diagnostic data is relevant to the timing of the problem by checking timestamps. Note that data from different sources can have different timestamp formats; be sure to understand the sequence of the different elements in each timestamp format so that you can tell when the different events occurred.

The specific method of analysis is unique to each data source, but one tip that is applicable to most traces and log files is to start by identifying the point in the data where the problem occurs. After you identify that point, you can work backward in time through the data to unravel the root cause of the problem.

If you are investigating a problem for which you have comparative data for a working and non-working environment, start by comparing the operating system and product configuration details for each environment.

You can use the collector tool to gather information about your WebSphere Application Server installation. The preferred method of analyzing the output from the collector tool is to send it to IBM Support for analysis. However, you can use the following topic to understand the content of the JAR file if you perform your own analysis: