Adaption is good!
Posted: Sat Mar 08, 2014 3:10 pm
Hi all,
Reading about DB2 internals and performance metrics has always been my favourite topic of interest. I recently read a column on Sequential Prefetch, one of the coolest features of DB2. DB2 can asynchronously read consecutive pages of data, if DB2 thinks you are likely to read pages sequentially. With a pre-fetch size of n pages, once you read (n/2)th page, DB2 triggers the next pre-fetch. I also know, that the DB2 optimizer decides to use Sequential Pre-fetch at BIND time.
With DB2 V9, dynamic pre-fetch feature makes it possible to switch from indexed-access to sequential pre-fetch on-the-fly. There are no dumb questions, so I'd like to ask, does this access path change reflect in the PLAN_TABLE? As an application programmer, how would I detect, if dynamic pre-fetch was leveraged?
Thanks,
Quasar C.
Reading about DB2 internals and performance metrics has always been my favourite topic of interest. I recently read a column on Sequential Prefetch, one of the coolest features of DB2. DB2 can asynchronously read consecutive pages of data, if DB2 thinks you are likely to read pages sequentially. With a pre-fetch size of n pages, once you read (n/2)th page, DB2 triggers the next pre-fetch. I also know, that the DB2 optimizer decides to use Sequential Pre-fetch at BIND time.
With DB2 V9, dynamic pre-fetch feature makes it possible to switch from indexed-access to sequential pre-fetch on-the-fly. There are no dumb questions, so I'd like to ask, does this access path change reflect in the PLAN_TABLE? As an application programmer, how would I detect, if dynamic pre-fetch was leveraged?
Thanks,
Quasar C.