Chapter 4: History Files and Archiving › Set Up a History File Reduction Scheme

Set Up a History File Reduction Scheme

You create a scheme for reducing data in the parameters file. Add or modify history file reduction schemes with the Parameter Editor as described in the chapter “Parameter Editor Commands.”

The reduction scheme must specify the following information:

• History description name—A history description 1 to 20 character name.
• Schedule—Specifies the time periods of the daily data for which data is archived, thereby allowing a subset of each day's data to be archived.
• Holiday schedule—Specifies the time periods for which data is archived on holidays. Performance Agent uses this schedule when the archived data falls on a day specified in PSDC$DATABASE:PSDC$HOLIDAYS.TXT. You can edit this file using any text editor.
• Granularity—The time span an individual history file represents. Granularity must be daily, weekly, monthly, quarterly, or yearly. Smaller portions of the historical data can be deleted, moved, or manipulated with a smaller granularity.
• Periodicity—Specifies the calendar period (typical day, week, month, or quarter) into which the daily data for a longer span of time is summarized. For instance, archived data with monthly granularity and weekly periodicity produces typical week data for the month; a daily periodicity would produce a typical day for the month. The value of periodicity must be less than the value of granularity. Periodicity must be daily, weekly, monthly, or quarterly.
• No Periodicity—Causes each day to be maintained individually. In effect, the value of periodicity is equal to the value of granularity.
• Interval of data resolution—The size, in minutes, of each interval record. Any even division of two minutes into a day is valid.
• Retention—The number of days Performance Agent keeps the history file beyond the time span it represents. The archive command first deletes history files that reach retention threshold and then archives the data. A history file representing a period of time before the retention period is subsequently deleted by the next archiving run. Therefore, to archive old daily files, extend the retention period to encompass their time frames.
• Model data switch—When enabled, causes Performance Agent to archive additional data, permitting subsequent modeling from the history file, and reclassification of the workload. All unique PIDs, user names, image names, account names, process names, UICs, and any combination of these names are kept separate.
• List of families—The name of the user group families by which process and image records are summarized if the model data switch is not enabled.

The daily Performance Agent data files and the archived data files reside in the database disk and directory as specified for the node in the CPD collection definition. A unique file exists for each node and for each unit of time defined by the granularity.

The size of a history file depends on the data reduction scheme. The most important element is the size interval you define. For example, if the interval is 60 minutes, the number of data records in the file is significantly smaller than if the interval were four minutes.

Also, if you reduce your history files according to a previously defined workload, the attributes of the workload definition affect the size of the resultant history file. For instance, if the workload definition was created with /UNIQUE_BY=image or /UNIQUE_BY=user qualifiers, your history files are larger because there is more data in a workload consisting of unique images or users.