Pick the unit you can forecast: hosts (Datadog, Dynatrace, Splunk), data volume in GB (New Relic, Elastic, Coralogix, SigNoz), events (Honeycomb) or seats. Bills from per-host products spike with container churn, and bills from per-GB products spike with noisy logs. Grafana, Sentry and SigNoz also exist as self-hosted software, which is a separate operating and support decision from their paid cloud plans.
Infrastructure Pro $15 per host per month billed annually ($18 on-demand). Infrastructure Enterprise $23 ($27 on-demand). APM from $31 per host. Logs from about $0.10 per GB.
Annual-billing price for Infrastructure Pro. On-demand is $18. Each product (APM, logs, RUM, security) is priced separately and usage-based.
When to choose it
Teams that want infrastructure monitoring, APM, logs and security from one vendor and accept a catalog of separately priced products.
What to confirm
Pricing is per product and per unit (hosts, GB, spans, custom metrics), so the headline $15 per host covers infrastructure only and a full stack adds APM, logs and more.
Free forever tier. Pro from $19 per month plus usage. Enterprise starts at a $25,000 per year spend commit.
This is Grafana Cloud (managed). Pro is a platform fee plus usage. Self-managed Grafana OSS (grafana.com/oss/grafana) is a separate product you run yourself, free to download, with Enterprise features requiring a license.
When to choose it
Teams that want open-source-compatible dashboards and metrics, logs and traces as a managed service with a generous free tier, or already run Prometheus and Loki.
What to confirm
Pro starts at $19 per month plus usage, and Enterprise starts at a $25,000 annual spend commit. Self-managed Grafana OSS is a different, free product that you run and support yourself.
Documented capabilities
Free forever tier for personal projects and early startups
Pro plan with usage-based metrics, logs and traces and 13-month metric retention
Host, GB ingested, event and user billing each break differently when you add containers, turn on debug logs or hire engineers, so the cheapest list price is often not the cheapest bill.
Export a month of host counts, log GB per day and trace volume and apply each vendor's meter to it.
Retention and rehydration cost
Low ingest prices can hide extra charges for retention beyond a default window or for searching archived data.
Find the default retention for logs, metrics and traces on the plan you would buy and what extending it costs.
Cardinality and sampling controls
High-cardinality metrics and unsampled traces are the usual source of surprise bills, and controls to cap them differ by vendor.
Ask for the vendor's guidance and tools for limiting custom metrics, indexed logs and trace sampling, and test them on a staging service.
OpenTelemetry and portability
Native OTLP ingest and open standards reduce lock-in and make a second-vendor pilot cheap.
Send the same OTLP telemetry from one service to two vendors and compare setup effort and dashboards.
Self-hosted versus managed
Open-source builds shift scaling, upgrades and on-call to your team and often lack the vendor's billing, support and newest features.
Read the self-hosted docs for stated support, upgrade and feature limits and estimate the engineer time to run it.