Driver 01
Unpredictable billing
Invoices that jump with custom metric cardinality and log volume.
Migration playbook: Datadog → Prometheus and Grafana
Datadog to Prometheus and Grafana migration
Replace a growing Datadog bill with OpenTelemetry, Prometheus and Grafana, keeping the dashboards and alerts your on-call team relies on.
Migration drivers
Why teams make this move
Driver 02
Proprietary agents
Instrumentation tied to vendor agents instead of vendor-neutral OpenTelemetry.
Driver 03
Retention cost
Paying a premium to keep traces and logs for more than a few weeks.
Execution sequence
How the migration runs
Each phase ends with a check you can verify — data parity, error rates, latency — and the rollback path is agreed before any traffic moves.
- 01Phase
OpenTelemetry Collector
Deploying the OpenTelemetry Collector to receive traces, metrics and logs in OTLP.
- 02Phase
Metrics backend
Setting up Prometheus or VictoriaMetrics with long-term storage on object storage.
- 03Phase
Dashboards and alert rules
Translating Datadog dashboards and monitors into Grafana dashboards and Alertmanager rules.
- 04Phase
Datadog agent removal
Removing the Datadog agents once alert coverage has been checked.
Risk prevention
Pitfalls that derail this migration
Risk 01
High-cardinality labels
Exporting user IDs or transaction hashes as Prometheus labels, which exhausts memory.
Risk 02
Copied alert thresholds
Porting Datadog anomaly monitors to PromQL without retuning thresholds and evaluation intervals.
Risk 03
Log shipping bandwidth
Sending uncompressed logs over public networks instead of aggregating them locally first.
Before and after
What we measure
We take a baseline before any change and report the same numbers after cutover, from your own tools. They are the evidence of whether the migration worked — not figures promised in advance.
- Monthly cost
- Datadog invoice vs the new stack's infrastructure and licenses
- Alert coverage
- Datadog monitors with a working equivalent before agents go
- Retention
- Days of metrics, logs and traces kept, by data type
Questions
Frequently asked migration questions
No. OpenTelemetry with Grafana Tempo or Jaeger gives you end-to-end traces and waterfall views. The UI differs, and some Datadog-specific features need a replacement or a decision to drop them.
Yes. Grafana Alerting and Prometheus Alertmanager integrate with PagerDuty, Opsgenie, Slack and custom webhooks.
Rehearse the cutover before the real one
Tell us about your data volume, traffic and timeline. An engineer will reply within one business day to set up a call about the migration plan and its rollback path.