first of all - please, excuse me for my bad English... :-(
I'm currently using DB2 v8.1 and trying to design a small business-database.
The issue, that drives me mad, is to keep rolling through the most current 36 months of data. I guess, that using partitioned tablespace based on date columns (in this case, months like "102003") is the solution for this problem. But I fail to describe it in DDL. :-(
DB2 offers partitioning with DB2 EEE V7 or DB2 ESE V8 (with partitioning option). However, this is hash partitioning (data randomly spread across all partitions) and not range partitioning.
Hash partitioning is for improving performance with parallel operation on a multi-CPU and/or multi-node system. This is usually employed on very large databases that frequently do tablespace scans.
Range partitioning is usually for ease of administration, especially when the range is by year, month, etc, although there are some performance benefits also. But this feature is not yet available on DB2 for Unix, Linux, and Windows (but is available on DB2 for OS/390). IBM has said that it will be available in a future release (date unknown).