Datadog guide · Cost and sizing
Datadog pricing explained: what drives the bill, and how to estimate it before you ask for a quote
Datadog prices each product separately, on its own unit: hosts for infrastructure and APM, gigabytes and millions of events for logs, test runs for synthetic tests, sessions for real user monitoring. Its public pricing page lists an annual-commitment rate and a higher on-demand rate for most of them. Most estimates go wrong on the units, not the rates: hosts nobody counted, tags that multiply metrics, and logs switched on by accident. This guide explains how each unit is counted and gives you an inventory to fill in before anyone quotes you.
Published 7 min read
In short
- Each product has its own unit. A Datadog bill is the sum of several meters, so estimate each product separately.
- Hosts are counted hourly and billed on the highest hourly count after the busiest 1% of hours is set aside.
- Allowances are pooled across your account. Each host includes a number of containers and custom metrics; usage above the pooled allowance is billed.
- Four common surprises: cloud instances counted as hosts without an Agent, an Agent inside every container, high-cardinality tags, and log volume nobody planned.
The unit behind each product
| Product | Billing unit | How it is counted |
|---|---|---|
| Infrastructure monitoring | Per host, per month | A host is a physical or virtual operating-system instance, metered every hour |
| APM (distributed tracing) | Per APM host, per month, plus span volume | A host counts as an APM host when an application on it sends traces; ingested spans are counted in gigabytes and indexed spans by number, above an allowance per APM host |
| Log management | Per gigabyte ingested, and per million log events indexed | Ingestion and indexing are separate meters; indexed logs also carry a retention period |
| Custom metrics | Allowance per host, then billed above it | Each unique combination of metric name and tag values counts as one custom metric |
| Containers | Allowance per host, then by container hours | Counted on the fractional hours each monitored container runs |
| Database monitoring | Per database host, plus normalised queries | Unique database hosts monitored each hour, and the number of normalised queries tracked |
| Synthetic tests | Per 10,000 API test runs; per 1,000 browser test runs | A test runs from every location you select, at every scheduled interval |
| Real user monitoring | Per 1,000 sessions | A session ends after 15 minutes of inactivity or 4 hours of continuous activity |
| Network monitoring | Per host, plus flows | Hosts monitored, and the traffic flows between source and destination |
On the public pricing page on 2 November 2026, the Infrastructure Pro plan included 5 containers and 100 custom metrics per host, and the Enterprise plan 10 containers and 200 custom metrics per host. These allowances are counted across your whole account, not per host. APM included 150 GB of ingested spans and 1 million indexed spans per APM host per month, also summed across all APM hosts.
How hosts are counted, and why short spikes rarely set the bill
Datadog meters hosts every hour. At the end of the month it sets aside the busiest 1% of hours and bills on the highest count in the remaining 99%.
In a 30-day month (720 hours), 1% is about 7 hours. A scale-out that lasts less than about 7 hours in the month does not set the bill; one that runs for a day does. If your clusters scale up for month-end processing, a campaign or Ramadan traffic, count the hosts you run during that peak, not your average.
Annual commitment, on-demand, and usage above commitment
- Two rates. Datadog's pricing page lists an annual-commitment rate and a higher on-demand rate for most products.
- Committing to a minimum. Datadog's billing documentation describes a plan in which you commit to a monthly minimum and host hours above it are billed at an hourly rate.
- Logs above commitment. For log products, the pricing page states that usage above the monthly commitment is billed at the annual price plus 50%.
- Retention. On-demand log retention defaults to 15 days; longer retention for indexed logs is part of what you choose and pay for.
- Discounts. The pricing page mentions multi-year and volume discounts; their terms are set in each contract.
Four things that make estimates wrong
1. Cloud instances counted without an Agent. Datadog bills every EC2 instance that its AWS integration picks up as a host, whether or not an Agent runs on it. It does not bill twice when both are present, provided it can match them; on instances that require IMDSv2, Datadog documents that this needs Agent 7.64.0 or later, or a setting that prefers IMDSv2. Use the integration's tag filters to exclude instances you do not want monitored.
Read more
2. An Agent inside every container. If the Agent is installed in each container, each container is counted as a host. Datadog recommends one containerised Agent per host instead.
3. Tags that multiply metrics. A custom metric is counted once per unique combination of tag values. Datadog's own example: one latency metric sent from two hosts, tagged by endpoint and status, produces four custom metrics. A histogram multiplies that by five. Add a tag with thousands of values, such as a user or order ID, and the count grows with it. Keep identifiers like these out of metric tags; put them in traces or logs where you need them.
4. Logs nobody planned. Debug logs left on after a release, or a new log source added without a review, raise ingestion, and indexing too if those logs are indexed. Datadog's Agent ships with OpenTelemetry log ingestion switched off by default, which it explains is to prevent unexpected log billing; switch it on deliberately. For how to keep log cost under control, see What drives your Datadog log bill.
Estimate before you ask for a quote
Fill in this inventory for each environment (production, staging, test). It turns "how much will Datadog cost?" into quantities anyone can price.
| What to count | Where to find it | Unit it drives |
|---|---|---|
| Servers and virtual machines, at your monthly peak | Virtualisation and cloud consoles; CMDB | Infrastructure hosts |
| Cloud instances the cloud integration will discover | Cloud console, with the tags you plan to filter on | Infrastructure hosts |
| Hosts running services you will trace | Service catalogue; deployment manifests | APM hosts |
| Containers per host, at peak | Kubernetes or container platform metrics | Containers above the allowance |
| Custom metrics: metric names × tag combinations | Existing metrics pipelines; planned dashboards | Custom metrics above the allowance |
| Log volume per day, and what must stay searchable for how long | Current log platform or log files; retention policy | GB ingested; events indexed; retention |
| Database hosts and their engines | DBA inventory | Database monitoring hosts |
| Synthetic tests × frequency × locations | Your list of critical journeys | API and browser test runs |
| Web and mobile sessions per month | RUM pilot, or a sample, rather than web-analytics session counts | RUM sessions |
Worked example for synthetic tests (arithmetic only):
one API test every minute from 3 locations runs 60 × 24 × 30 × 3 = 129,600 times in a 30-day month, about 13 units of 10,000 runs. One browser test every 5 minutes from 2 locations runs 288 × 30 × 2 = 17,280 times, about 17 units of 1,000 runs. Frequency and locations multiply each other, so set them per journey, not by default.
Watch the estimate after go-live
- Estimated usage metrics. Datadog publishes near real-time usage estimates for most products, which you can chart and alert on. Datadog states they differ from billable usage by 10–20% on average, more for small usage, so treat them as an early warning, not as the invoice.
- The usage pages. Hourly usage and subscription details are in Datadog's plan and usage pages, which Datadog limits to the Admin role and to roles granted billing permissions. Decide who owns that review, and how often.
- A monthly check. Compare hosts, custom metrics, log GB and indexed events against the inventory above, and find the source of any jump before the next invoice.
Questions to settle with whoever quotes you
- Which plan tier for each product, since the tier changes the container and custom-metric allowances?
- Committed quantities per product, and how usage above them is billed?
- Retention for indexed logs, and whether long-term retention is included?
- Which cloud integrations will run, with which tag filters?
- How are synthetic test runs counted when a test runs from several locations?
- Currency, invoicing and payment terms. Datadog's public prices are listed in dollars.
How Interkey helps
Interkey is an official Datadog partner and implements and operates Datadog for organisations in Saudi Arabia. Cost design is part of our implementation, not a later rescue project: we size the estate with the inventory above, set tagging standards and log pipelines before data starts flowing, and on existing deployments start with an audit of what is collected, what it costs and which alerts fire.
Questions buyers ask
Is there a free tier?
Datadog's pricing page lists a free infrastructure tier for up to five hosts with one day of metric retention. It suits a trial, not production monitoring.
Why does our bill not match the usage dashboard?
Datadog's estimated usage metrics are calculated in near real time and differ from billable usage by 10–20% on average. The invoice uses the billing method for each product, such as the 99th-percentile host count.
Do short autoscaling peaks raise the host bill?
Only if they last long enough. The busiest 1% of hours in the month, about 7 hours in a 30-day month, is set aside before the host count is taken.
Are containers billed as hosts?
Not when one containerised Agent per host monitors them: each host includes a container allowance, and containers above it are billed by the hour. If an Agent is installed inside each container, each container is counted as a host.
Where can we see current prices?
On Datadog's official pricing page. Prices, allowances and plan names change, so use the live page rather than figures copied into third-party articles.
Next
Related guides
Next step
Get your estate sized before you budget
Tell us roughly how many servers, containers and services you run, your daily log volume and which products you are considering. A specialist from the Riyadh team replies on Saudi working days, in Arabic or English.
Or directly
+966-11-2180999 [email protected]Tawuniya Towers, North Tower, 7th Floor, King Fahad Highway, Olaya, P.O. Box 56835, Riyadh 11564, Saudi Arabia