On Mar 31, 2008, at 8:23 PM, Ravi Chemudugunta wrote:
>> In general I would recommend that you benchmark them using
>> as-close-to-real load as possible again as-real-as-possible data.
>
> I am running a benchmark with around 900,000 odd records (real-load on
> the live machine :o ) ... should show hopefully some good benchmarking
> results for the two methods.
Please do, and please share. I know the docs say that exception
blocks make things "significantly" more expensive, but I think that
the community also sometimes loses the forest for the tree. Setting
up a savepoint (AFAIK that's the actual expense in the exception
block) is fairly CPU-intensive, but it's not common for a database
server to be CPU-bound, even for OLTP. You're usually still waiting
on disk.
--
Decibel!, aka Jim C. Nasby, Database Architect decibel@decibel.org
Give your computer some brain candy! www.distributed.net Team #1828