Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Mailing Lists -> Oracle-L -> RE: Timed_statistics, True or False
On 21 Aug 2000, at 17:22, Nguyen, Long wrote:
> I agree with Winnie that timed_statistics should be turned on. In fact I
> think Oracle should just turn on statistics by default and not making the
> timed_statistics parameter available. I think Oracle has made many init.ora
> parameters available which on one hand giving us DBAs the power to
> manipulate the system performance and on the other hand making the system
> more complex than necessary and making our job harder.
>
> Long
On 21 Aug 2000, at 17:22, Nguyen, Long wrote:
> I agree with Winnie that timed_statistics should be turned on. In
fact I
> think Oracle should just turn on statistics by default and not
making the
> timed_statistics parameter available. I think Oracle has made
many init.ora
> parameters available which on one hand giving us DBAs the
power to
> manipulate the system performance and on the other hand
making the system
> more complex than necessary and making our job harder.
>
> Long
I'm afraid I have to disagree with you on this. I like having as many parameters as possbible available for tuning, switching and general tweaking. A good example of why this is important is this specific parameter - TIMED_STATISTICS. Due to a bug in 8.1.5 it causes cursor sharing to stop (and not cursor caching as my disobedient fingers typed before). What this means is that if you can't turn that parameter off because it has been put out of reach you will either have to 1) Live with the high parse to execute ratio and associated poor performance or 2) Get a patch to repair the problem.
The huge number of tunable parameters is what makes Oracle rather daunting to get started with, but they are like the sails of a ship - each one has its purpose, sets of them work together, a master sailor can trim them to capture every available breath of wind and a lubber can set them to carry away in a storm or capsize the whole dang ship. Received on Tue Aug 22 2000 - 15:24:57 CDT