The number of page faults can remain fairly stable and not affect performance significantly. However, once the number of page faults reaches a certain threshold, our system will be quickly and severely degraded. This is even more evident for HDDs, but it affects solid-state drives (SSDs) as well.

The way to ensure that we don't run into problems regarding page faults is to always have a staging environment that is identical to our production in setup. This environment can be used to stress test how many page faults our system can handle, without deteriorating performance. Comparing the actual number of page faults in our production system with the maximum number of page faults that we calculated from our staging system, we can find out how much leeway we have left.

Another way to view page faults is via the shell, looking at the extra_info field of the serverStatus output:

> db.adminCommand({"serverStatus" : 1})['extra_info']
{ "note" : "fields vary by platform", "page_faults" : 3465 }

As the note states, these fields may not be present in every platform.