MongoDB provides detailed statistics at the collection level via its collStats and top commands. The Compass Performance tab will even show you the Hottest Collections:
However, since it is not uncommon to have multiple application teams sharing a MongoDB cluster, it is often desirable to see this type of information rolled up to the database level.
The analyze-db.py script uses information from collStats and top to roll up information on db cache use and a list of hottest dbs:
$ python3 analyze-db.py
MongoDB Database Stats
Analyzing Database Cache Use...
Database Cache Used (MB)
-------- ---------------
local 1238.762
demos 109.216
sample_airbnb 97.617
sample_mflix 39.312
sample_training 26.328
sample_weatherdata 11.996
-------- ---------------
Total 1523.747
Calculating Hottest DBs...
Database CPU CPU Read CPU Write
-------- --------- --------- ---------
demos 6.11% 5.77% 0.34%
local 0.41% 0.41% 0.0%
sample_mflix 0.26% 0.0% 0.0%
enron 0.14% 0.0% 0.0%
-------- --------- --------- ---------
Total 6.92% 6.18% 0.34%
Analysis Complete
Note the script also persists its findings back to a dbstats database, which allows for historical analysis. For example, you can use MongoDB Charts to graph the results over time:
