Last July 2019, I was able to attend the ContainerCamp conference in Sydney courtesy of DevOpsGirls. Here are some of my notes from the talk Keeping an Eye on Serverless Containers.
What is Observability?
- an attribute of the system/server
- allows you to pinpoint problems of the system/server without remoting into it directly
- a timestamped record of discrete events
- forensics information read to analyse a failure mode
- actionable metrics, closely related to user experience
- manage known failure modes of the system
- profile and monitor microservices via request tracing
- pinpoint where failures occur and what causes poor performance
How to be Observable?
- Instrumentation - writing the code to generate telemetry
- Ingestion & aggregation - consume/collect the telemetry
- Visualisation & interrogation - looking at the telemetry (and understanding it!)
- Custom instrumentation for each app
- Transporting telemetry data
- Dependency on third-party libraries