Parallel Seeding of Data into TIBCO Patterns - Search

To improve the performance when seeding data into TIBCO Patterns - Search, index entities (specified in the IndexerConfig.xml file) are processed in batches instead of processing sequentially.

You can specify the batch size in the Configurator by using the Records per batch for the advance matching engine (com.tibco.cim.netrics.records.batchSize) configuration property. 

The AsyncIndexingEntities class is added for the parallel processing of index entities. A queue message is sent to the Async queue. Based on the listeners, messages are picked and processed.

Based on the batch size, records are picked and indexed in separate threads. You can see the batch number and number of records processed in the elink.log file.

For normalized entities, parent repositories are indexed first, and then child records are indexed in parallel.