Practice vocabulary for metrics instrumentation: counters, histograms, gauges, Prometheus client libraries, and cardinality management.
0 / 5 completed
1 / 5
A metric that only goes up and tracks how many times something has happened (e.g., total HTTP requests) is called:
The counter increments on each request — counters only go up (or reset to zero on restart). Use rate() in PromQL to compute per-second rates from counters.
2 / 5
A metric that samples and bins numeric observations to show statistical distribution (e.g., request latencies in p50/p95/p99) is called:
The histogram tracks request duration distribution — histograms bucket observations and allow calculating percentiles (e.g., histogram_quantile(0.95, ...) in Prometheus).
3 / 5
A metric that represents a current value that can go up or down (e.g., number of active connections right now) is called:
The gauge shows current active connections — unlike counters, gauges can decrease. Examples: active goroutines, memory usage, current queue depth.
4 / 5
The standard way to add metrics collection to an application is described as:
We instrument the application with Prometheus client library — Prometheus has official client libraries for Go, Python, Java, Ruby, and others that expose metrics via HTTP.
5 / 5
When too many unique label combinations are used in a metric (e.g., user_id as a label), causing memory and performance problems, this is called:
The metric cardinality explosion vocabulary — cardinality = number of unique time series. Using high-cardinality labels like user_id or request_id creates millions of series, overwhelming Prometheus.