I would like to count unique users by day, week, and month. I'm not really sure what's the preferred Splunk method to do this. I've been experimenting with different searches and summary indexes/accelerated reports. I'm struggling with determining the most efficient solution.
I believe populating a summary index with a daily search such as
search | sistats dc(user_id) BY field1, field2, field3
will work. However, my concern is that since there are many unique users the index can become quite large. Changing the search to something like
search | stats count BY user_id, field1, field2, field3
is another option but this would create many events which I don't think would be any more efficient than just searching the source events.
Are there any recommended solutions for counting distinct users over periods of time?
... View more