Re: Oracle Application Partitioning

Hrishy wrote:

>What is meant by oracle Application Partitioning esp
>in the case of RAC ?

Generally people mean that you should try to make different nodes work on different sets of data (to minimise contention).

For example:

- allocate different functions on different nodes (eg sales on RAC1, purchasing on RAC2); you may be able to do this simply by mapping users to specific nodes based on their departments
- allocate work geographically - eg NY staff work on NY node, DC staff on DC node
- split data horizontally - use RAC1 for new business, RAC2 for account maintenance (ie dealing with rows further back in the table)

There are lots of potential ways to split your application - and which is best depends on the circumstances. No method is completely foolproof.

One thing I have campaigned for in the past is to make sure your application servers don't screw up your partitioning strategy by thoughtless load balancing. See http://preferisco.blogspot.com/2006/04/micro-partitioning-pooled-sessions-in.html. The idea is to maximise partitioning, but accept that failover could also lead to an increase in contention on remaining nodes.

Regards Nigel

--
http://www.freelists.org/webpage/oracle-l