Monitoring
When designing a software system, we undertake many explicit and implicit assumptions. We always try to make the best decisions based on our knowledge but there may be some parameters that we didn't take into account or just underestimated.
Using monitoring, we can validate our assumptions, verify that our application performs as intended, and scales as expected. Good monitoring systems are also vital for detecting software bugs and help us detect early potential security incidents.
What should we monitor?
By far the most important metric to monitor in MongoDB is memory usage. MongoDB (and every database system for what it's worth) uses system memory extensively to increase performance. No matter if we use MMAPv1 or WiredTiger storage, used memory is the first thing we should keep our eyes on.
Understanding how computer memory works can help us evaluate metrics from our monitoring system. These are the most important concepts related to computer memory.
Page faults
RAM is expensive...
