Skip to content

TELEMETRY EVIDENCE

Prometheus Add time-series metrics to connected evidence.

Prometheus already holds the performance signals your teams trust. AutoObserve ingests or queries that metrics evidence and connects it to logs, traces and change context — without forcing a rip-and-replace.

Evidence
Time-series metrics
Protocol
Prometheus remote write / scrape
Enables
Performance analysis
Status
GA

Read metrics docs

Prometheus

  1. Metrics

  2. AutoObserve

  3. Performance context

  4. Investigation

WHY THIS INTEGRATION MATTERS

Your metrics stay useful when they join the full evidence model.

PromQL dashboards answer performance questions. AutoObserve connects those answers to logs, traces and deployments so investigations do not stop at a single graph.

  • Coexistence

    Keep Prometheus and Grafana — AutoObserve subscribes to metrics evidence in parallel.

  • Performance analysis

    Latency and saturation signals feed connected production intelligence.

  • Cross-signal correlation

    Metrics spikes align with logs, traces and change events automatically.

  • Multi-DSL access

    Query Prometheus alongside other backends inside one investigation plan.

EVIDENCE RECEIVED

Time-series metrics from Prometheus

Counter, gauge and histogram series become performance evidence inside AutoObserve — linked to identity and change context.

  • Time-series metrics
  1. Ingest

    Remote write or scrape paths deliver Prometheus series to AutoObserve.

  2. Normalize

    Align metric labels with service identity and resource attributes.

  3. Connect

    Join metrics to logs, traces and deployments in one window.

  4. Investigate

    Use performance evidence inside evidence-backed RCA.

HOW AUTOOBSERVE USES THIS EVIDENCE

Metrics become connected performance evidence.

Prometheus is not replaced — it is connected. AutoObserve treats your existing metrics as one evidence class in a multi-signal investigation.

Provides

  1. Time-series metrics

AutoObserve creates

  1. Performance analysis
  2. Saturation signals
  3. Cross-signal correlation
  4. Investigation queries

Prometheus is performance evidence. AutoObserve is connected understanding.

COEXIST WITH PROMETHEUS

Add AutoObserve without replacing your metrics stack.

Remote write, federation and sidecar paths keep Prometheus authoritative while AutoObserve builds connected evidence.

Forward selected series to AutoObserve via remote_write.

  1. Prometheus

  2. Remote write

  3. AutoObserve

QUICK START

Connect Prometheus metrics evidence

Configure remote write or scrape forwarding, then verify checkout-api latency series appear in connected evidence.

  1. Configure remote write

    Add an AutoObserve remote_write endpoint to prometheus.yml for production series.

    remote_write:
      - url: https://${AUTOOBSERVE_HOST}/api/v1/write
        bearer_token: ${AUTOOBSERVE_API_KEY}

    Connect metrics →

  2. Align metric labels

    Ensure service and deployment labels match OpenTelemetry identity conventions.

    Resource attributes →

  3. Verify performance evidence

    Confirm checkout-api latency metrics appear during a test investigation window.

    Verify telemetry →

WHAT SUCCESS LOOKS LIKE

Metrics connected. Performance evidence ready.

Success means latency and saturation series correlate with service identity — not merely that remote write is configured.

  • Series ingesting

    checkout-api latency metrics arrive within expected scrape intervals.

  • Labels aligned

    service and deployment labels match telemetry identity.

  • Cross-signal correlation

    Metric spikes align with logs and traces in one window.

  • Investigation queries work

    PromQL-backed evidence appears in investigation plans.

REAL INCIDENT

Same checkout spike — performance metric in focus.

The canonical checkout-api degradation with the latency metric highlighted — the performance signal that triggered investigation.

  1. Metric

    14:31:00

    Checkout latency +184%

Walk through the evidence chain · 1/6

Evidence step 1 of 6

TROUBLESHOOTING

Common Prometheus integration failures

  • Problem

    No metrics arriving

    Likely cause

    Remote write URL, auth token or relabel rules dropping series.

    Resolution

    Verify remote_write config, bearer token, and that target series match relabel keep rules.

    Verify

    checkout-api latency series visible within two scrape intervals.

  • Problem

    Metrics present but identity missing

    Likely cause

    service label does not match OpenTelemetry service.name.

    Resolution

    Align Prometheus target labels with resource attribute conventions.

    Verify

    Metric series resolve to checkout-api in production context.

  • Problem

    High-cardinality series dropped

    Likely cause

    Aggressive relabel or cardinality limits on ingest path.

    Resolution

    Review label set and drop rules; keep service-level aggregates.

    Verify

    Primary latency histogram available for investigation queries.

PRODUCTION INTELLIGENCE

How Prometheus contributes to production understanding

Metrics evidence enables performance investigation inside connected production intelligence.

EvidenceEnables
Time-series metricsPerformance analysis
Saturation signalsCapacity reasoning
HistogramsLatency distribution

DESIGN PARTNERS

Validate this evidence source with your production stack.

Design partners connect real evidence sources and validate investigation workflows before GA — with direct engineering access.