In some cases, such as for full-text index partitions,
a tradeoff exists between indexing performance and content-based retrieval
(CBR) query performance.
Query optimization: Index partitions
Index partitioning is the grouping of object index information
into separate indexes based on object property values. You can improve
content-based retrieval (CBR) query performance by using index partition
properties in the search conditions of your query.
Query optimization: Multiple search servers
Configure multiple IBM Content Search Services search
servers to distribute the load and allow more full-text indexes to
be searched in parallel. Dedicate each server to either Index or Search mode
as opposed to the dual IndexAndSearch mode. Multiple
search servers are especially helpful if your content-based retrieval
(CBR) queries typically search many full-text indexes.
Query optimization: Allocate heap memory for concurrent queries
Search servers create temporary in-memory objects for a
query to facilitate the running of that query. To optimize CBR query
performance for queries that are run concurrently, allocate sufficient
heap memory for the in-memory objects.
Query optimization: Allocate heap memory and increase the timeout interval for searchables
Search servers create and use in-memory representations
of full-text indexes to facilitate the running of queries. These in-memory
representations are known as searchables. To optimize CBR query performance,
configure servers to create searchables infrequently. Also, allocate
sufficient heap memory for searchables.