| Provider | $/mo estimate ▼ | $/host/mo | Logs $/GB | Pricing model | Free tier | Self-host | Trustpilot | Notes |
|---|---|---|---|---|---|---|---|---|
| Loading… | ||||||||
Why does the monitoring bill run away?
Two compounding drivers: per-host price multiplies with every new server/container (many in microservices), and log ingestion is billed per gigabyte — a busy app can produce hundreds of GB of logs per month. On top come separately APM, RUM (user tracking), synthetic tests, user licenses and long retention. Many are surprised when the free 100 GB runs out and the bill multiplies.
How to control: set sampling and retention limits for logs, route "noise" to cheaper storage, and consider a self-hosted open stack (Grafana + Loki + Mimir + Tempo, or SigNoz) — then you pay only for infra, not per-GB. Also compare uptime monitoring (availability only, cheaper) and block storage (log archive).
How to choose?
For small teams usage-based (New Relic 100 GB/mo free, Grafana Cloud free tier, Better Stack) are safe — you pay only when you grow. Per-host pricing (Datadog, Dynatrace, Splunk) is predictable with few servers but expensive with many containers. Cheapest log ingestion is Datadog/Honeycomb/Better Stack ($0.10/GB); clearly most expensive is Azure Monitor ($2.30/GB).
Data ownership and avoiding hidden costs: SigNoz and Grafana are open source — self-hosted you pay only for servers, which is overwhelmingly cheap at large data volume (and keeps data in the EU). Hyperscaler-native (CloudWatch, Google Cloud Monitoring, Azure Monitor) integrate tightly with their cloud but custom metrics and log GBs inflate the bill. Always check retention: free 14–15 days is enough for debugging, longer retention costs.