New Relic vs Datadog
New Relic MCP and Datadog MCP both turn an observability platform into an investigation tool for an agent, and both are official remote servers — but they reflect each platform's query model. New Relic's server centers on NRQL: an agent can convert natural language to NRQL, execute it against NRDB, fetch entities and their golden metrics and related entities, read dashboards, and triage by listing alert conditions, policies, recent issues, and searching incidents. Datadog's server leans into its broad signal coverage: search and analyze logs, query metrics and their metadata, pull full APM traces and spans, list hosts and services and their dependencies, search RUM events, and inspect monitors and incidents the way an on-call engineer would. The deciding question is which platform holds your telemetry and whether you prefer New Relic's NRQL-and-entity model or Datadog's signal-by-signal investigation surface. Here is a balanced look.
How they compare
| Dimension | New Relic | Datadog |
|---|---|---|
| Query model | NRQL-first: natural_language_to_nrql plus execute_nrql_query against NRDB, with an entity model (golden metrics, related entities) layered on top. | Signal-first: dedicated tools to search/analyze logs, query metrics, pull traces and spans, and search RUM, hosts, and services. |
| Tracing and APM | Surfaced through entities and NRQL queries against the unified NRDB rather than dedicated trace/span tools. | First-class trace tooling — get_datadog_trace and search_datadog_spans — plus service and service-dependency search for APM-style investigation. |
| Incident triage | list_alert_conditions, list_alert_policies, list_recent_issues, and search_incident bring New Relic alerting and issues to the agent. | search_datadog_monitors, search_datadog_incidents, and get_datadog_incident let the agent inspect monitors and walk incident detail. |
| Real-user monitoring | Available through entities and NRQL against browser/mobile data in NRDB, not a standalone RUM-event search tool. | Dedicated search_datadog_rum_events tool for querying real-user-monitoring events directly. |
| Best-fit task | Asking questions in plain English that become NRQL, exploring entities and golden metrics, and triaging issues across a New Relic account. | On-call-style investigation that pivots across logs, metrics, traces, RUM, hosts, and services, then jumps into monitors and incidents. |
Verdict
Pick by where your telemetry lives and how you like to investigate. Reach for New Relic MCP when your observability is on New Relic and you want a natural-language-to-NRQL workflow over NRDB with an entity-and-golden-metrics model for triage. Reach for Datadog MCP when your data is in Datadog and you want to investigate like an on-call engineer — pivoting across logs, metrics, APM traces and spans, RUM events, hosts, and services, then into monitors and incidents. Both are official remote servers, so the real decision is platform plus whether you prefer NRQL-and-entities or a signal-by-signal surface. In short: New Relic for NRQL-driven, entity-centric analysis; Datadog for broad, signal-pivoting incident investigation.
FAQ
- Do I need NRQL knowledge to use the New Relic server?
- Not necessarily — the server includes a natural_language_to_nrql tool that translates a plain-English question into NRQL, which it then executes against NRDB. You can also supply NRQL directly via execute_nrql_query if you prefer.
- Which has stronger tracing support?
- Datadog exposes dedicated trace and span tools (get_datadog_trace, search_datadog_spans) plus service-dependency search, making APM-style trace investigation first-class. New Relic surfaces traces through its entity model and NRQL against NRDB rather than standalone trace tools.