A ClickHouse specialist vs a full-stack observability platform
Datadog is a mature, full-stack observability platform — APM, tracing, logs, on-call — with a ClickHouse integration bolted on. chmonitor only monitors ClickHouse, but goes far deeper: no agent to install, ClickHouse-native pages for merges/parts/replication, a recommend-only query/MV advisor, and a built-in AI agent. Neither replaces the other everywhere — here's the honest comparison.
chmonitor vs Datadog
Datadog legitimately wins on full-stack APM and on-call — that's not what chmonitor is for.
| chmonitor | Datadog | |
|---|---|---|
| Setup time | Point at your ClickHouse host — Docker, Helm, or Cloudflare Workers, ~5 min | Install the Datadog Agent and ClickHouse integration check on every node |
| Agents required | None — reads system tables directly over the ClickHouse protocol | A Datadog Agent process runs on every monitored host |
| ClickHouse-specific depth | Purpose-built pages for query log, merges, parts, replication queue and Keeper quorum | A fixed ClickHouse integration metric set; no dedicated merge/part/replication drill-down UI |
| Full-stack APM, distributed tracing & log management | ClickHouse monitoring only, by design | Industry-leading APM, tracing and log management across your entire stack |
| On-call, escalation & incident management | No on-call/escalation product | Mature on-call routing, escalation policies and incident management |
| AI agent for ClickHouse system tables | Chat agent + MCP server with tools scoped to ClickHouse's own system tables | Datadog's AI features reason over Datadog telemetry, not ClickHouse system-table semantics specifically |
| Query advisor (recommend-only) | Ranked index/rewrite suggestions with EXPLAIN-based impact estimates; never auto-applies DDL | Not part of a general infra-monitoring integration |
| Materialized view / projection designer | Recommend-only MV & projection suggestions derived from real query patterns | Out of scope for the integration |
| Anomaly detection | Statistical baselines scoped to ClickHouse cluster metrics, built in | Mature, general-purpose anomaly and forecast monitors across any metric you send it |
| Alerting & webhooks | ClickHouse-specific thresholds and outbound webhooks run everywhere — narrower by design; alert history and inbound-event storage persist with a D1 binding (hosted cloud or a Cloudflare Workers self-host) | Broad monitor types, deep notification integrations and escalation across your whole stack |
| Cost | Free self-hosted (GPL-3.0); hosted cloud in early access | Infrastructure Monitoring ~$15–$23/host/month; APM adds ~$31/host/month; logs billed per GB (2026 list pricing, varies by commitment) |
| Open source / self-host | GPL-3.0 — Docker, Kubernetes (Helm), Cloudflare Workers | SaaS only, no self-hosted option |
No agents, no per-host bill, for ClickHouse specifically
If ClickHouse monitoring is the whole job, an agent-based, per-host-billed platform is a lot of surface area for one system.
Nothing to deploy per node
chmonitor reads ClickHouse system tables directly. No Datadog Agent process to install, update or babysit on every host.
Merge/part/replication depth
Dedicated pages for the parts of ClickHouse a general infra integration doesn't drill into.
Advisor + AI agent included
A recommend-only query and MV/projection advisor plus a built-in AI agent, scoped to ClickHouse — not a general infra add-on.
Free self-hosted, period
GPL-3.0, no per-host bill. Datadog Infrastructure Monitoring runs ~$15–$23/host/month before APM or logs.
More comparisons: chmonitor vs Grafana · chmonitor vs ClickHouse Cloud · ClickHouse vs TimescaleDB · ClickHouse vs Postgres · ClickHouse vs Druid vs Pinot
Start monitoring — advisor and alerts built in
Open the hosted dashboard and connect a cluster, or self-host with Docker, Kubernetes, or from source. Self-hosting is always free.