Summarizing a database

Visualizer lets you graph very detailed performance information, down to the workload and device level. While this level of detail may be appropriate for a week or two, it isn't suitable for a long-term database. Visualizer includes a database summarize function that replaces sets of values with single values representing the average, minimum, peak, or Global Max value.

Average, minimum, and peak summaries are created for each object in the data. This means that you could end up with some surprising results, especially in the case of peak summaries.

For example, a peak summary of daily data contains the peaks for each object in the data. Those peaks could occur at a variety of intervals during the day. If you looked at the total for all peaks for a logical system, the total could add up to more than the utilization for that logical system.

Global Max takes a different approach to summarizing data. When you specify Global Max, Visualizer finds the interval when cumulative CPU utilization (by logical system) reaches a maximum for each physical system. The Global Max summary contains all values for all objects during the peak interval by physical system.

Global Max values apply to the following platforms:

• MVS measurements
• VM measurements
• iSeries measurements
• Oracle measurements
• Sybase measurements

Summarize also lets you merge the results of the Summarize event with an existing summary database. This lets you gradually accumulate long-term summary data (such as monthly summaries) from partial data sources (such as daily measurement data).

Using Summarize by shift

If you want to display data points for summarized shift data, you must create a unique tag for each shift to be summarized. For example, if you summarize a week's worth of data for three shifts, creating unique tags for each shift, the resulting graph displays 21 data points (three shifts times seven days).

Interpreting summarized data

When you summarize a database, you add new records to the database. Visualizer generates the new summary record by computing averages, minimums, or peaks (maximums) from a selected set of records that are already in the database. As part of the Summarize event, you can delete measurement records from which the summary values are computed.

When analyzing peak or minimum summary data, the resulting composite data, such as sums and averages, might be wrong. Visualizer calculates peaks and minimums for each individual data item in the database. The peaks and minimums don't necessarily represent the same measurement interval as the values with which they are combined. Composite data from Average summaries, however, are accurate in all graphs.

Redirecting summarized data

To redirect summarized data from multiple measurement databases, click keep data for existing summary. Doing so ensures that Visualizer doesn't delete (by overwriting) existing summaries in the target database.

When you redirect summarized data, Visualizer doesn't delete existing summaries in the target database, even if they have the same summary tag and date as the new summary. This lets you summarize data from multiple source databases into the same target database without overwriting one another.

We recommend redirecting summarized data with the same intervals. You get conflicting data if you redirect summary data with the same tag and date range, but different time intervals, for example, 15 minute intervals as well as 45 minute intervals. Some data from the original summary, which should not be in the new one, may remain in the target database. This causes inaccurate reporting.