Question: Is it a good idea to leave the search performance objects enabled 24/7?

Answer: It depends on your use cases. If you’re attempting to collect data pertaining to a problem that occurs sporadically, and you’ve chosen configuration values that don’t introduce a painful amount of performance overhead, there’s no reason you can’t keep the objects enabled on an ongoing basis.

Question: What kind of performance impact will enabling the search performance objects have?

Answer: Performance overhead, in terms of CPU and memory usage, should be negligible when using the DataStax Enterprise’s default configuration values. However, the overhead introduced by enabling the objects varies as the configuration is modified (described in the following sections). For instance, setting longer TTLs and shorter refresh intervals leads to higher memory and disk consumption.

Question: Should I enable the search performance objects on every node in my cluster?

Answer: The search performance objects should only be enabled on search nodes, that is, nodes where indexes reside that can observe search operations. While it is perfectly acceptable to enable the objects across an entire cluster, enabling them on a single node for observation first is a good way to mitigate risk.

Question: Can I use existing tables with secondary indexes on some columns, and create search indexes on other columns in the same table?

Answer: Do not mix search indexes with secondary indexes. Attempting to use both indexes on the same table is not supported.