Tips to Improve Performance of Introspection Implementation
To improve introspection performance, you can:
• Put the TDV server on a machine with many CPUs.
• Increase the Connection Pool Maximum Size setting of the data source (be aware that the physical data source can get impacted if this number is too high). To change this setting, open the data source, and modify settings on the Advanced tab.
• Set the introspect.override_parallelism_degree in the data source capabilities file. The data source capabilities files for all supported data sources can be found at:
<TDV_install_dir>\apps\dlm
Setting introspect.override_parallelism_degree in the data source capabilities file to a number higher than the number of CPUs might degrade performance. However, if there is network latency between the TDV and data source, setting this to a number higher than the number of CPUs might keep the TDV busy doing what it can while it waits for responses from the data source. This number cannot be higher than the maximum pool size of the data source. For example, this is a good option if you do not want to change the current connection pool size, but want to consume three-quarters of the pool instead of half of it.
• Specify introspect.max_batch_size in the data source capabilities file to tune introspection batching to match the batching capabilities of the data source. (Batching is supported for DB2, Oracle, PostgresSQL, Teradata, and Oracle E-Business Suite data sources. The default max_batch_size is 999.)