Re: cloned database slower

Hi Susan,

Your problem can have many causes. Here are a few of the more obvious ones :

1. Even though the 2 instances have the same buffer cache size, the buffer cache is (almost) empty at startup. The original database has run the query for several times, so the blocks retrieved by the query are much more likely to be in the buffer cache on the original DB than on the clone DB => quicker response time on the original DB. To chech that, make conditions equal for both DB's by either
2. Bouncing both of them, if possible.
3. Pinning the objects involved in the query in the buffer cache on both instances.
4. Are you sure the databases really have the same instance parameters ? You have copied the init.ora file, but many init.ora-parameters are DYNAMIC (especially in 9i), which means that the original DB could have changed parameters without that being reflected in the init.ora file. To check that, make the following query on each DB

    select * from v$parameter;     

and compare the results. Any deviations should be examined.

3) Do you have the same session parameters on both queries? If you have something like OPTIMIZER_GOAL = FIRST_ROWS on the first one and not the second one, it could easily explain the whole deviation.

4) The O/S may, for some O/S reasons, give more ressources to the original DB.

• Kenneth Koenraadt

susana73_at_hotmail.com (Susan Lam) wrote in message news:<7186ed56.0212231104.50360850_at_posting.google.com>...
> I've copied/cloned a database to another unix filesystem on the SAME
> server. I did it by copying datafiles over then reconstruct control
> files. Now the user is complaining the cloned database is
> significantly slower than original database! We have a complicated
> query it takes 1 hour in original db but 3 hours on cloned database.
>
> The init.ora parameters are exactly the same except database name,
> location of archive, control files... The nature of filesystem is
> exactly the same. And they are in the same Sun Solaris server. The
> CPU is pretty much at the same state when we did the comparison test.
>
> Any idea? Will cost base stat copy over after cloning?
>
> Thanks,
> Susan
Received on Thu Dec 26 2002 - 06:29:12 CST