Hi splunkers,
Problem:
We have quite big database with events ( ~3 millions events / month), so search works not that fast. For instance, a search for 6 month takes ~10 minutes. The thing is that in most cases, that's not necessary to process each event, so we came up with a possible solution.
Possible solution
We can speed up search by processing only each Nth (10th or 100th) event. For rough results that would be enough. The question is how to (is that possible at all) make such search?
i would search for events where date_second is equal to some given second. You can even do more combinations like (date_minute<30 date_second=11) OR (date_minute>=30 date_second=47).
This is also an area of ongoing development in the product.. Check with your SE in the future.
There is no row number in Splunk data so, taking the Nth would not be possible/efficient. I would go with David's answer to apply filter based on extracted fields, easiest ones are the date_* fields.
i would search for events where date_second is equal to some given second. You can even do more combinations like (date_minute<30 date_second=11) OR (date_minute>=30 date_second=47).
This is also an area of ongoing development in the product.. Check with your SE in the future.
Thanks a lot, I don't know why such a nice idea didn't visit me 🙂