Datadog vs Grafana Cloud vs New Relic — which observability platform?

Blockers

• ! Dashboard JSON export via API — No bulk historical data export — metrics, logs, traces stored in Datadog are not bulk-exportable; no API for raw telemetry dump — Rating: none
• ! PromQL remote_read API (metrics), LogQL API (logs), Jaeger-compatible query API (traces), Dashboard JSON API — All signals exportable via open standard APIs — Rating: full
• ! Historical Data Export via NerdGraph to S3 bucket — Up to 200M data-point rows per export; time range must end 12h+ in past; result files expire after 1 week. Also: Streaming Data Export via Kinesis Firehose using NRQL rules — Rating: partial
• ! DogStatsD proprietary metric submission; Datadog-specific semantic conventions/extensions added to OTel ingestion; DD-native SDK instrumentation requires rework for vendor-neutral; Intelligent Retention, Metrics without Limits are proprietary controls
• ! NRQL is proprietary query language (SQL-like but NR-specific); dashboards and alerts built on NRQL require translation if migrating; query and alerting layer proprietary despite first-class OTel ingestion
• ! Re-instrument and forward to new backend; OTel Collector community Datadog receiver as migration path — Cannot backfill historical data from Datadog; DD receiver coverage not feature-complete; requires switching from DD-native SDKs to OTel — Pain: high
• ! Re-instrument and dual-ship via OTel Collector — Cannot backfill historical data; DD-specific semantic conventions need rework — Pain: high
• ! Export via open APIs (PromQL remote_read, LogQL, Jaeger API) + self-host open-source stack — All protocols are open standards; no re-instrumentation needed — Pain: low