Python for DevOps

Chapter 7. Monitoring and Logging

When Noah worked in startups in San Francisco, he used his lunch break to exercise. He would play basketball, run up to Coit Tower, or practice Brazilian Jiu-Jitsu. Most of the startups Noah worked at would have a catered lunch.

He discovered a very unusual pattern coming back from lunch. There was never anything unhealthy left to eat. The leftovers were often full salads, fruit, vegetables, or healthy lean meats. The hordes of startup workers ate all of the unhealthy options while he was exercising, leaving zero temptation to eat bad food. There is something to be said for not following the crowd.

Likewise, the easy path is to ignore operations when developing machine learning models, mobile apps, and web apps. Ignoring operations is so typical it is like eating the chips, soda, and ice cream at the catered lunch. Being normal isn’t necessarily preferred, though. In this chapter, the “salad and lean meats” approach to software development is described.

Key Concepts in Building Reliable Systems

Having built companies for a while, it’s fun to look at what worked on the software engineering portion versus what didn’t. One of the best antipatterns out there is “trust me.” Any sane DevOps professional will not trust humans. They are flawed, make emotional mistakes, and can destroy entire companies on a whim. Especially if they are the company’s founder.

Instead of a hierarchy based on complete nonsense, a better approach to building reliable systems ...