Implementing Modern DevOps

Whitebox monitoring

Whitebox monitoring, as the name indicates, is the monitoring that needs to know about how the system is built in order to raise alerts on certain events happening inside of our application or infrastructure. These metrics are quite fine-grained (unlike blackbox monitoring), and once we have been alerted, they are the answer to the main questions of a postmortem analysis:

• Where is the problem?
• What is causing the problem?
• Which flows are affected?
• What can we do to avoid this in future?

Among other metrics, these are a fairly interesting set of examples:

• Function execution time
• Errors per time unit
• Requests per time unit
• Memory usage
• CPU usage
• Hard drive usage
• I/O operations per time unit

As you can see, there is an ...