Defining, Generating, and Punching a DMCL › Journal Buffers and Journal Files › Sizing Journal Files

Sizing Journal Files

Disk Journal Attributes

When you define disk journal files consider the following topics:

• How many disk journal files to define
• The number of blocks in each disk journal

Number of Disk Journals

For optimal journal processing, you should define at least three disk journal files. When one file is full, CA IDMS/DB can immediately write to another file. While CA IDMS/DB writes to the alternate file, you can offload the full disk journal file to an archive file using the ARCHIVE JOURNAL utility statement. If CA IDMS/DB fills the second file, it can swap to a third file even if the ARCHIVE JOURNAL utility is still offloading the first file.

Batch Update Jobs May Require Added Files

You may need to increase the number of disk journal files when you run a batch program that updates a large volume of data. An added disk journal can prevent a situation in which the offload utility fails to complete its task before the remaining disk journal files fill.

Place Files to Avoid Offload Contention

To reduce contention during offload operations, you should:

• Place the disk journals on disk packs that do not contain database or dictionary files
• Assign each disk journal to a different volume and channel

Disk Journal File Size

The size of a disk journal file affects:

• How often the disk journal gets offloaded to the archive journal and the amount of time required to accomplish the offload. A small disk journal size means a greater number of archive tapes to keep track of since the last database backup. A large disk journal size means CA IDMS/DB will need more time to offload the disk journal to an archive file.
• The risk of losing data due to an I/O error on a journal file. A smaller file reduces the potential data loss while a larger one increases it.
• The amount of time required to perform a warmstart following a system failure. If the disk journal files are large, it may take longer for CA IDMS/DB to read through the journal in use at the time of the system failure.

  Note: You can enhance warmstart performance by using the FRAGMENT INTERVAL options of the SYSTEM system generation statement or of the DCMT VARY JOURNAL command.