Re: Local managed tablespace better for performance ?

Andreas Wizemann wrote:
>
> Hi,
> i'm considering using local managed tablespaces to speed up a large
> batch application. Is there an 'easy' way to check, if this will give
> me the expected performance boost ? Means, can i select some values or
> ratios from my running system where i can expect to get some benefit ?
> Like checking cache hit ratio before increasing buffer...
>
> Or, which are the 3 most efficient parameters to tune for a job
> inserting about 200.000 rows ?
> eg. initrans, freelist, cache, rbs,...
>
> DB is 8.1.7.1 on aix 4.3.3-ML08 using IBM ESS SHARK as DASD device.
>
> regards
> Andreas
> Andreas Wizemann

Switching to locally managed tablespaces has it's good and bad points. Before making this change, you will need to investigate the performance of dictionary managed vs. locally managed.

As far as the 3 most efficient parameters...it depends on what the problem is. If you are having a wait event that points to freelist contention, then that may be a place to look. I will venture to say there are other methods you could use to speed up the insert of 200,000 rows...like use SQL*Loader, or array inserts. It's really hard to generalize without knowing what the specifi issues are.

Good luck.

-- 
Ron Reidy
Oracle DBA
Reidy Consulting, L.L.C.