Re: When to use partitioned tables ?

There are several reasons for partitioning, not limited to the following three:

1. Size & related performance - there is no definite size when you should start partitioning. If performance seems to be suffering and the typical query could benefit from up front separation, then consider it. For example, if you can isolate the data by geography or date range and most queries look at that isolation level, then partitioning might be able to reduce the effective data set size simply by throwing away the partitions that are not needed.
2. Administration - you want to administer the table in pieces, perhaps to keep the underlying table up while maintaining indexes or whatever on a segment of the table.
3. Data management - especially useful in date partitioning; quickly add the latest data and/or quickly throw away the oldest data without dramatically impacting your users

/Hans

Francesco Marchioni wrote:

> Hi all,
> I'm studying how to partition a table. I wonder what are
> the standards about this: for example how many records
> you need to have in your Table in order to start thinking
> about partitioning ? I speak about an ordinary sized table...
> I just want an advice if 10.000 records are worth partitioning
> or you should have about 1.000.000 in order to outweigh the
> cost of partitioning (I think that there must be some overhead
> in your query when you use a partitioned a table)
> Thanks
> Francesco
Received on Tue Jul 03 2001 - 01:23:28 CDT