Tuning Guide
Also available as:


Indexing is primarily IO driven. Tuning indexing tends to focus on the search index (Solr or Elasticsearch). Problems with indexing running too slow will often manifest as Kafka not commiting in time. This results from the indexing backing up so that it fails batches and the poll interval in Kafka is exceeded. The issue is actually with the index rather than Kafka.

You can modify many performance tuning properties for indexing using Ambari or Storm Flux. Modifying properties using Ambari is simple and can be performed by any user. However, you should have knowledge of Storm Flux usage and formatting before attempting to modify any Flux files.

The indexing properties materialize as follows:

Ambari UI -> properties file -> Flux -> Storm