PMM adds and adds, more and more metrics and collectors for them but there is neither design nor arch for it.
It is kinda assumed that there are some monitoring standards (like Prometheus) that PMM follows but they are either not defined or not followed (maybe obsolete). For example `collect` parameters for the exporter.
This leads to many issues and performance degradation of the systems under monitoring.
We also don't have plan about better observability strategy (matching metrics relations, analyzing them and better future advisors based on that).
Goal of this Epic is to gather information, define requirements and start formulating some problem statement for the metrics collections, metrics analyzes and etc.
Phase 1 limited to collectors, flexibility of gathering metrics and configuring collectors and define set of standards and interfaces.
Also to set a base for arch, design and problem statement docs for the next phases.
We should take community first approach defined in our objectives and goals: