Chapter 4: History Files and Archiving › Use History Files

Use History Files

You can use history files as the data source for generating performance evaluation reports, analysis reports without factory rules, dump reports, models, and graphs. To specify a particular history file, use the /HISTORY_DESCRIPTOR qualifier when generating reports or models.

History files cannot be used to generate an analysis report from the Factory rules. Other rules, normally evaluated at analysis report time, are evaluated at archive time. This means you can generate an analysis report from history data using auxiliary rules.

The performance evaluation report output is different in that it includes a new section on the archived rule occurrences. This section includes the following data for each rule that had an occurrence:

• The rule identifier
• The number of rule occurrences
• The history interval time of the first and last rule occurrence
• The archiving process, for each history file descriptor found in the parameters file, reads the daily data file and create, update, or supersede the history file for each node. For a given history descriptor, the Periodicity attribute determines how the archiving process deals with daily data.

If no value is specified for Periodicity, each set of daily data is processed and added to the history file. Created to cover a specific time span, the history file is then updated with each succeeding set of daily data. In this case, updating the history file does not change the version number of the file.

If you specify a value for Periodicity in the history descriptor, daily data is used in calculating new values to supersede the data stored in the history file. For each set of daily data processed, archiving creates a new version of the history file for that time span. In this case, use the Parameter File Editor to limit the number of file versions you want to retain.

If you archive using the two default history file descriptors, MONTHLY_MODEL and MONTHLY_USER, then the archiving process summarizes data into history files according to each history definition. So it summarizes data into intervals representing one hour, 24-hours per day, and create one history file per calendar month for each history definition.

The file name is in the following format:

path:PSDC$nodename_dd-mmm-yyyy_dd-mmm-yyyy.name

path

Represents the directory specified in the collection definition.

nodename

Is the name of the node from which the daily data was archived.

dd-mmm-yyyy

Is the date representing the start date of the historical data file.

dd-mmm-yyyy

Is the date representing the end date of the historical data file.

name

Is the unique history file descriptor name identifying the database specification in the parameters file.

If the archiving run is terminated prematurely, a subsequent archiving or reporting command employs the journal file. The most recent version of the history file is used to produce a valid history file to update or to create the requested report.

While updating the history files, the archiving process creates a journal file for each history file being updated. This process names each journal file to correspond with the history file using a file type of .name_JOU. If the history file is invalid, the archiving process uses the journal file and the invalid file to recover.

No journal files are created for history files that have a history file descriptor periodicity attribute. In this case, the original history file is retained and no new version is created. The journal files are automatically deleted when no longer needed.