I have read the doc re. using these parameters and one should specify the tablespaces' extent size value for BLOCKSIZE (assuming that all tablespaces using the same bufferpool have the same extentsize value).
I am looking for real-life experiences in DW environment where many table scans are being done (and much sequential prefetching) who have seen performance benefits (or even issues) when using NUMBLOCKPAGES and BLOCKSIZE.
We're at UDB ESE 8.2.7 (aka FP14) on AIX 5.3 ML04.

Much obliged,