Re: How to re-org an empty file

Rob -

IMO, It depends on what works for you. We have a set of conversion files for an insurance application which is being migrated away from the iSeries. Many of them end up containing millions of records at the end of a conversion run.

There is a significant amount of system overhead involved in performing file extents, so I reviewed the number of records in each file after the most recent conversion and did a CHGPF FILE(XXXX) ALLOCATE(*YES) on each and sized them such that the initial allocation [ SIZE() ] is the number of current records + ~ 10 %.

Example:
The address file had 450,000 records, so I changed it like this:
CHGPF FILE(ADDRESS) ALLOCATE(*YES) SIZE(500000 10000 1000)

These files all get either reorg'ed or cleared (or both) before each conversion run. [This activates the space allocation...]

After making these changes, one job which previously took 9 hours to run took around 7 hours, and the only real variable which changed was the file allocation changes.

- sjl


Rob wrote:

Interesting on the allocated size. Why would one use that - to ensure
that you had the space available when you ran the process that goes to
maximum size? Or, you just knew you were going to grow to that and you
wanted to avoid extents? I guess IBM gives us the tools to do what we
want and if we misuse them then that's our choice. Preallocating space
for maximum size on all tables when they're never ALL going to grow to
maximum size is probably a real disk chomper.