Related: How might an admin find out if they have too few, or too many, unicorns for their web traffic?
Related: How might an admin find out if they have much more RAM than they need?
My understanding is that the number of unicorns is generally scaled according to the number of CPUs, but it needn’t be so - if, for example, the host config has changed since setup.
The amount of RAM should scale according to how much is needed. I don’t know if a huge import of many many posts and users necessarily means more RAM is needed - the question would be, how much of the database is accessed to serve each web request, or each regular task?
For me, a 10 min burst of maximum CPU use, against a background of very low use, would not indicate a problem. (It would be a problem if it meant a very busy and time-critical forum, such as say a sports/gambling forum, had some delay to serving pages at times of peak interest. But my forum is not time-critical or performance-critical.)
I don’t know how one could tabulate the range of recent response times, but that might be a good bit of data to look at. Either from the web-serving side or from the database side.