Grafana Cloud vs Datadog for Metrics: Free Tier Limits, Retention, and Real Costs

You set up a metrics pipeline, pick what looks like the most generous free tier, and three weeks later you're staring at a bill you didn't expect. Both Grafana Cloud and Datadog are capable platforms, but their pricing structures are almost opposites — and the one that looks cheaper on the landing page is not always the one that stays cheap once your team grows.

This article cuts through the marketing copy and compares what you actually get on each platform, where the costs spike, and which one makes more sense depending on how you work.

What you'll learn

• Exactly what each free tier includes for metrics, hosts, and retention
• How each platform's pricing model scales (and where it gets painful)
• Retention windows and what happens to your data after they expire
• Hidden costs that don't show up in the headline pricing
• Which platform fits which team size and use case

The Two Pricing Philosophies

Datadog charges per host. Every infrastructure agent you run is a billable unit, and the per-host fee covers a bundle of features. Grafana Cloud charges by consumption — metrics series, logs ingested, traces, and so on. Neither model is inherently better, but they reward different usage patterns in very different ways.

A team with five servers but thousands of custom metrics will hit Datadog's per-host fee quickly but benefit from the bundled features. The same team on Grafana Cloud might stay in the free tier longer, then face a cost spike when their active time series count crosses the threshold. Understanding which axis your usage grows along is the first question to answer.

Grafana Cloud Free Tier: What You Actually Get

Grafana Cloud's free tier is genuinely useful for small projects. At the time of writing, the free plan includes 10,000 active metrics series, 50 GB of logs, 50 GB of traces, and 500 VUh of frontend session data per month. Dashboards and alerting are included without a host count limit.

The metrics retention on the free tier is 13 months for Prometheus-compatible metrics. That is longer than what many paid tiers on competing platforms offer, and it means you can actually do year-over-year comparisons without upgrading. Logs retention is 30 days on the free tier.

Where Grafana Cloud's Free Tier Breaks Down

Ten thousand active series sounds like a lot. It isn't if you have dynamic labels. A Kubernetes cluster with pod-level labels can generate tens of thousands of series almost immediately, because each unique combination of label values creates a new series. If you instrument a microservice with a high-cardinality label like user_id or request_id, you'll burn through the free tier in hours, not months.

The other constraint is that some integrations and data sources are restricted or unavailable on the free plan. If you need Grafana OnCall for incident management or Grafana SLO tracking at scale, you'll need to upgrade.

Datadog Free Tier: What You Actually Get

Datadog's free plan is more constrained. You get up to five hosts, one-day metric retention, and access to a limited set of integrations. Logs, APM, and most of the interesting features require a paid plan. The free tier is essentially a trial, not a long-term option for production use.

One-day retention is the real killer. You can't use it for trend analysis, capacity planning, or debugging anything that happened yesterday morning. For any serious use, you're looking at a paid plan almost immediately.

Datadog's Paid Tiers and Host Pricing

Datadog's Infrastructure Pro plan (the most common entry point) is billed per host per month. The per-host fee covers infrastructure metrics with a 15-month retention window, a set of built-in integrations, and dashboards. APM, logs, and custom metrics are all separate line items.

Custom metrics are where Datadog budgets get complicated. Each host includes a fixed number of custom metrics in the bundle. Anything beyond that is charged per metric per month. If your application emits a lot of business metrics — order counts, queue depths, API latency histograms — those can accumulate fast and add a meaningful amount to your monthly bill.

Retention: A Closer Look

Retention matters more than most teams realize when they're choosing a platform. You need historical data to answer questions like