Only single thread spawned querying Aggregate stream udf from Java client


While performing load test on Aerospike we observed that Java Client uses only one connection to query the Aerospike Server.

Java client is calling a Stream UDF lua function to aggregate. Performance is significantly slow compared to write.

Even though we have specified 2000 connection per node, max connection created while writing is around 120 and only 1 connection while reading.

Are we missing any configuration?

For a query, a connection is made to each node in the cluster. The query can’t be parallelized any further than per node.

© 2015 Copyright Aerospike, Inc. | All rights reserved. Creators of the Aerospike Database.