Are ad-hoc Splunk searches multi threaded?

[ppuru]

In simple words, if I execute an ad-hoc search on a Splunk environment that has multiple CPU cores across indexer clusters and search peers, will my search use a single core or will it use more that that? Is there a .conf file that can define this?

[skoelpin · Accepted solution]

By default, each search will use a single CPU core. Luckily this is configurable and you can tell Splunk how many searches to dedicate per CPU core. Assuming your running on bare metal, you can hyperthread your CPU to get more out of it. This singe search per core limitation may get in the way when scaling because you may want to accelerate data models which use 3 concurrent CPU cores to rebuild the summary range every 5 minutes.

Before adding more CPU to your search heads, you should add indexers as this will increase search speeds and "lock" CPU for less time when searching (i.e. faster search speeds leads to less utilized CPU cores). You also have the option of adding faster storage to existing indexers which will greatly increase IOPS, thus increasing search speed and using less CPU.

https://conf.splunk.com/files/2016/slides/jiffy-lube-quick-tune-up-for-your-splunk-environment.pdf

[richgalloway]

Every search, whether ad-hoc or scheduled, uses a single core on the originating search head and one on each indexer.