Metrics & Timeseries

Metrics & Timeseries

Metrics are numbers over time — CPU usage, response times, request counts. In Dynatrace, the timeseries command is how you query them.

The timeseries Command

// Basic: average CPU across all hosts
timeseries avg(dt.host.cpu.usage)

// Split by host
timeseries avg(dt.host.cpu.usage), by:{dt.entity.host}

// Multiple metrics at once
timeseries {
  cpu = avg(dt.host.cpu.usage),
  mem = avg(dt.host.memory.usage)
}, by:{dt.entity.host}

// Custom time range
timeseries avg(dt.host.cpu.usage), by:{dt.entity.host}, from:now()-2h

Common Metrics

Metric                                What It Measures              Unit
────────────────────────────────────  ──────────────────────────    ──────────
dt.host.cpu.usage                     Host CPU utilization          Percent
dt.host.memory.usage                  Host memory utilization       Percent
dt.host.disk.used.percent             Disk usage                    Percent
dt.service.request.response_time      Service response time         Microseconds
dt.service.request.count              Request throughput             Count
dt.service.request.failure_count      Failed requests               Count
dt.service.request.failure_rate       Failure percentage             Percent

Aggregations

timeseries avg(dt.host.cpu.usage), by:{dt.entity.host}       // Average
timeseries max(dt.host.cpu.usage), by:{dt.entity.host}       // Peak
timeseries percentile(dt.service.request.response_time, 95)  // P95
timeseries sum(dt.service.request.count)                      // Total

Filtering Inside timeseries

// filter:{} is MORE PERFORMANT than | filter (filters at data source level)
timeseries avg(dt.host.cpu.usage),
  by:{dt.entity.host},
  filter:{dt.entity.host == "HOST-ABC123"}