Observability

Factors such as distributed architectures, elasticity and transience of resources, increased transactional depths of workloads, and high cardinality of metadata makes canonical monitoring insufficient in meeting feedback and speed requirements of DevOps.

Instrumenting traditional agents that enable monitoring is super useful. This can be further augmented by instrumenting tracing libraries within code during development, to enable spans for distributed tracing. They provide the context to improve observability during runtime. Zipkin and Jaegar are a few open-source tools that support a variety of technologies, like Brave for Java from Zipkin.

Observability driven development enables developers to 'observe' the breadcrumbs built in the code, helps debug and crash MTTR during fault resolution.

Building diagnostics and visualization on the Bigdata of traces for near real-time feedback is a key focus area. This would take Observability, DevOps, and SRE capabilities to a higher level.