Do unused search-time field extractions significantly impact performance if the extracted fields are not used?

[wegscd]

I have a new analyst requesting to add some search-time field extractions for sourcetype=syslog to simplify reporting for a subset of syslog events. I'm concerned that running all the extra extracts for searches that will not use the extracted fields will negatively impact performance for the rest of my user population.

Does Splunk optimize out unused extractions? If not, is there a good way to quantify how expensive this is gonna be?

[sundareshr]

One way to test is by changing the Search Mode. There are 3 options...

• Fast mode speeds up searches by limiting the types of data returned by the search.
• Verbose mode returns as much event information as possible, at the expense of slower search performance.
• Smart mode, the default setting, toggles search behavior based on whether your search contains transforming commands. For transforming searches, it behaves like Fast mode. For searches without transforming commands, it behaves like Verbose mode.

Run your search and look at the execution time in the Job Inspector (http://docs.splunk.com/Documentation/Splunk/6.1/Knowledge/ViewsearchjobpropertieswiththeJobinspector)

https://docs.splunk.com/Splexicon:Searchmode