For applications to be considered production-ready, they should include monitoring that actively collects performance metrics and raises alerts when things go wrong.

Not so long ago, applications were deployed on a shared, beefy hardware that was capable of running multiple applications and workloads in parallel. Software deployments were done manually by the operations guy according to instructions written by the developer. If the instructions were properly written, they also included a step to set up monitoring for the application. The entire process was carried out manually, which was fine when deployments only happened on a weekly, monthly, or quarterly basis.

But deployments started to be more frequent when virtualization technologies gained popularity, and it became easier to deploy each application on its own virtual machine to make it more resilient against issues that were common with applications deployed on the shared hardware.

The self-service...