Skip to content

Metric Types

PRISM supports various metric types to represent different kinds of measurements.

Available Metric Types

Type Constant Description Example
Coverage coverage Percentage of coverage Test coverage
Rate rate Frequency or percentage Error rate, availability
Latency latency Time duration P99 latency, MTTR
Ratio ratio Proportion Success ratio
Count count Absolute count Deployment count
Distribution distribution Statistical distribution Latency percentiles
Score score Composite score Health score

Coverage

Coverage metrics measure the extent of coverage for a control or practice.

Characteristics

  • Range: 0-100%
  • Trend direction: higher_better
  • Common unit: %

Examples

{
  "id": "ops-test-coverage",
  "name": "Test Coverage",
  "metricType": "coverage",
  "trendDirection": "higher_better",
  "unit": "%",
  "current": 85,
  "target": 95
}

Rate

Rate metrics measure frequency or percentage over time.

Characteristics

  • Can be percentages (0-100%) or counts per time period
  • Trend direction varies by metric
  • Common units: %, per hour, per day

Examples

{
  "id": "ops-availability",
  "name": "Service Availability",
  "metricType": "rate",
  "trendDirection": "higher_better",
  "unit": "%",
  "current": 99.95,
  "target": 99.99
}

{
  "id": "ops-error-rate",
  "name": "Error Rate",
  "metricType": "rate",
  "trendDirection": "lower_better",
  "unit": "%",
  "current": 0.1,
  "target": 0.01
}

Latency

Latency metrics measure time duration, typically for response times or recovery.

Characteristics

  • Trend direction: Usually lower_better
  • Common units: ms, seconds, minutes, hours, days

Examples

{
  "id": "ops-p99-latency",
  "name": "P99 Latency",
  "metricType": "latency",
  "trendDirection": "lower_better",
  "unit": "ms",
  "current": 250,
  "target": 200
}

{
  "id": "ops-mttr",
  "name": "Mean Time to Recovery",
  "metricType": "latency",
  "trendDirection": "lower_better",
  "unit": "hours",
  "current": 2,
  "target": 1
}

Ratio

Ratio metrics measure proportions, often as decimals or percentages.

Characteristics

  • Range: 0-1 (decimal) or 0-100 (percentage)
  • Trend direction varies
  • Common units: none, %

Examples

{
  "id": "ops-success-ratio",
  "name": "Request Success Ratio",
  "metricType": "ratio",
  "trendDirection": "higher_better",
  "current": 0.998,
  "target": 0.999
}

Count

Count metrics measure absolute numbers.

Characteristics

  • Integer values
  • Trend direction varies by metric
  • No unit or custom unit

Examples

{
  "id": "ops-deploy-count",
  "name": "Daily Deployments",
  "metricType": "count",
  "trendDirection": "higher_better",
  "unit": "/day",
  "current": 5,
  "target": 10
}

{
  "id": "ops-incident-count",
  "name": "Open Incidents",
  "metricType": "count",
  "trendDirection": "lower_better",
  "current": 3,
  "target": 0
}

Distribution

Distribution metrics capture statistical distributions, often with percentiles.

Characteristics

  • Multiple values (p50, p90, p99, etc.)
  • Typically used with dataPoints array
  • Trend direction usually lower_better for latency distributions

Examples

{
  "id": "ops-latency-distribution",
  "name": "Latency Distribution",
  "metricType": "distribution",
  "trendDirection": "lower_better",
  "unit": "ms",
  "dataPoints": [
    {"timestamp": "2024-01-01T00:00:00Z", "value": 50, "label": "p50"},
    {"timestamp": "2024-01-01T00:00:00Z", "value": 150, "label": "p90"},
    {"timestamp": "2024-01-01T00:00:00Z", "value": 250, "label": "p99"}
  ]
}

Score

Score metrics represent composite or aggregated scores.

Characteristics

  • Range: Typically 0-100 or 0-1
  • May be computed from other metrics
  • Trend direction: higher_better

Examples

{
  "id": "ops-health-score",
  "name": "Service Health Score",
  "metricType": "score",
  "trendDirection": "higher_better",
  "current": 85,
  "target": 95
}

Trend Directions

Direction Constant Description
Higher Better higher_better Higher values indicate improvement
Lower Better lower_better Lower values indicate improvement
Target Value target_value Target a specific value

Choosing Trend Direction

Metric Type Typical Direction
Coverage higher_better
Availability higher_better
Error Rate lower_better
Latency lower_better
MTTR lower_better
Deployment Frequency higher_better
Health Score higher_better