USE Method
For every resource, track Utilisation, Saturation and Errors — the resource-centric complement to request-centric monitoring.
Applied to each resource in turn: CPU, memory, disk I/O, network, connection pools, thread pools, queue capacity, file descriptors.
Utilisation is the fraction of time the resource is busy. Saturation is the amount of work queued and waiting for it — and it is the leading indicator, because saturation rises before utilisation reaches 100% and long before latency visibly degrades. Errors are the resource's own error counters, which are frequently the fastest route to a root cause and the least-consulted signal in most estates.
The resource most often left unmeasured is the connection pool, and pool saturation is one of the most common causes of an outage. Requests wait for a connection, latency climbs, and every application-level metric points at "the database is slow" when the database is idle.