Re: partitioning table

HI,
The rows don't get moved very frequently. I found thet the main problem is due to sub query's. where I have to do a select of a selct of a select.

 I suspect it is better not to use the auto row delete & reinsert, but to delete the record and re-insert it myself. otherwise I suspect things could get messy with replication. steve

Peter L <news_at_abc.co.uk> wrote:

> On Fri, 15 Jun 2001 18:02:16 +0800, E_at_E.COM (steve) wrote:
>
> >Hi,
> >yes I just found out.
> >
> >
> >I was under tha assumption that i could partition my data on a given
> >col,then if I changed that col oracle would automatically move the data
> >to another partition. But that is not the case.
> >
> You can do this with partitioned tables. You have to alter the table
> to enable row movement, the default is for it to be disabled.
> Obviously, if you are updating the key you want to partition on, this
> isn't going to be good for performance. The row has to be deleted,
> from the old partition, then inserted to the new one, whenever an
> update results in a row needing to move partitions.
>
>
> >
> >I want to devide the data up into descrete areas so that oracle knows
> >where to find all the col1=n and does nto check the full 10k records.
> >steve
> >
> >
> This sounds like range partitioning could help. As long as your query
> includes 'where col1 = n' then it will be possible to eliminate
> partitions that cannot hold that value of n. The results can be
> dramatic in terms of performance, but to be honest, I wouldn't know
> whether it was worth it for 10k rows. Perhaps there's something else
> that needs tuning.
Received on Sun Jun 17 2001 - 21:30:46 CDT