In the SRE book from Google, there are two types of monitoring defined:
- BlackBox monitoring
- WhiteBox monitoring
This is generally accepted by everyone, leading to a solid amount of tools that are clearly differentiated around whitebox and blackbox monitoring.
One of the best comparisons I've ever heard on whitebox versus blackbox monitoring is the diagnosis of a bone fracture. When you first go to the doctor, he/she only has access to your blackbox metrics:
- Does the area have any bump?
- Is it painful on movement?
Then, once the initial diagnoses has been pronounced, the next step is getting X-rays from the area. Now we can confirm whether the bone is broken and, if it is, what is the impact in the system. The X-rays are the WhiteBox monitoring that the doctor is using.