Some general guidelines that can improve processing speed include the following:
- Close queries as soon as possible.
- Commit transactions as soon as possible.
- Change to dirty read (DR) when doing read-only operations where reading uncommitted changes are acceptable (some report writing, for example).
- Balance work between the client CPU and the XDB Server CPU. For example, try to minimize ORDER BYs and aggregates. When writing COBOL programs, take advantage of the COBOL sort facility (which can process on the client's workstation) instead of using ORDER BY (which processes on the XDB Server).
- To improve batch processing speed, consider turning off auto commit and forward logging if the process can be easily re-run.
- When using embedded SQL in a program (any language), and you are executing many similar queries (same structure, but different values), use a PREPARE statement and multiple EXECUTE statements. For example, specify one PREPARE to set up the structure. Then EXECUTE repeatedly using different host variables for each.
