Network observability was designed to complement the existing metrics, logs, and tracing tools you may already be using. These approaches are each a valuable part of an observability strategy but overall, they are each dependent on application-level data and they require developer instrumentation to be truly useful.
Tracking the network behavior between processes in a distributed system has a number of distinct advantages that fills gaps left by other tools. Notably:
Cloud infrastructure visibility: Flowmill agents collect data from the operating system and notably, the network stack in Linux, so they can detect infrastructure problems affecting distributed services.
Automatic instrumentation and discovery: By collecting data at the network level using eBPF, Flowmill agents do not require any changes to application code or container images.