Our organization has one of our Data Model (DM) searches for ES regularly running over 200 seconds to complete. Soon another source will be added to the DM, so I have been looking for ways to reduce the runtime.
I came across a site that suggested the macros that build the CIM DM's could be faster by adding the sourcetype alongside the index in the search. My thought was, "Why stop there?" You could add the source too, as long as it doesn't iterate with a date or some random number scheme. And even then, with reasonable wildcarding at the end I believe there would be a performance improvement.
I was told that this effort is unnecessary, even though in my unofficial tests over the same period, I found my modified searches to nearly twice as fast. So aside from the additional effort to build those searches and maintain them, why is this unnecessary?