OLTP applications and Oracle RAC concerns

For Online Transaction Processing (OLTP) applications, including Sterling Order Management System Software, one common concern is high insertion rates and the effect on index maintenance. In high volume OLTP applications, index leaf blocks have to be maintained and passed among the multiple nodes to keep them all in sync. Generally, when new records, like orders, are being indexed, they are being written to the right most part of the index. In very high transaction volumes, concurrent insertions could wait while a similar request is handled by a different node. The index leaf block for the right most part of the index cannot be released to another node until the request is completed. This forces more sequenced rather than simultaneous processing and is likely to drag significantly on performance.

Another example is the frequency with which inventory records are being accessed and updated.

Research suggests that other OLTP app vendors are generally aware of these issues — some only certifying for a maximum number of nodes and other articles suggesting optimal node / CPU configurations for Oracle RAC.

Some industry literature suggest using hash partition or reverse indexes to reduce or eliminate contention to enable OLTP applications for RAC. What isn't stated is that these techniques can negatively affect application performance which could slow down query response times.

Check with Technical Support for the latest information.