Disagree. Unless you go full into saas, you need a team to manage the infra or manage the interface to the tooling that provides the infrastructure. ie. tsdb storage (prometheus, victoriametrics), olap (clickhouse, etc), and standards. The author only mentions Grafana. Grafana doesn't store metrics, its displays them (and can alert on them too) it relies on datasources for this information.
If you intend to give a holistic view of the infra and the metrics within it then you'd have to string together N team's datasources or run centralized (but distributed query interface - ie thanos, vmselect). That requires heavy tooling in any decent sized business and a team to shepherd it.
Yes dev teams should care about and be responsible for their metrics. They should also know how their metrics make their way into the tsdb. On top of that they should have alerting on said pipeline.
blinded•9mo ago
If you intend to give a holistic view of the infra and the metrics within it then you'd have to string together N team's datasources or run centralized (but distributed query interface - ie thanos, vmselect). That requires heavy tooling in any decent sized business and a team to shepherd it.
Yes dev teams should care about and be responsible for their metrics. They should also know how their metrics make their way into the tsdb. On top of that they should have alerting on said pipeline.