Datadog is hired to show how servers and services are doing and to alert people when they are not. The bill grows with the number of hosts and products, which makes it a common line to review. Replacing it is a real engineering job, so below we cover prices as published, the open-source option, and a plan that keeps on-call coverage intact.
Monitoring is installed by engineers who need it to work at 3 a.m., and after that it is wired into everything: agents on every server, alerts routed to phones, dashboards pinned on the incident channel. Nobody wants to be the one who removed something that was about to page them.
The bill tends to grow with the infrastructure, often faster than anyone reviews it. Hosts that were retired, test environments and extra products add up, and the invoice is read by finance while the system is understood by engineering.
That is why the first step is usually not a replacement. It is a joint look, finance with the person on call, at what is being monitored and whether it still needs to be.
Only alternatives we have checked. An open-source project is listed with the license its repository reports, on the date shown; one whose license could not be read as a single standard id is not offered as open source. Read the license guide before running any of them.
| Alternative | Kind | License | Read from |
|---|---|---|---|
| Grafana | Open source | AGPL-3.0 | Grafana repositoryChecked Oct 25, 2026 |
Each feature below is what the product's own page or README says, linked to where we read it on the date shown. Pros and cons are limited to what those pages state and to what the path itself changes. We do not rank products on quality.
| Plan | Price as published | Includes | Read from |
|---|---|---|---|
| Infrastructure Free | $0 | Infrastructure Monitoring for up to 5 hosts, 1-day metric retention | Datadog pricingChecked Oct 25, 2026 |
| Infrastructure Pro | $15 per host per month, billed annually | Infrastructure Monitoring | Datadog pricingChecked Oct 25, 2026 |
| Infrastructure Enterprise | $23 per host per month, billed annually | Infrastructure Monitoring | Datadog pricingChecked Oct 25, 2026 |
List prices as printed on Datadog's pricing page on the checked date, in US dollars and excluding tax. Per-host rates shown are for annual billing; the page prints higher on-demand rates. Prices may differ at checkout, and only Infrastructure Monitoring was read. Other products are priced separately.
Monitoring is the job where a cheaper bill is easiest to buy and hardest to verify. The cost of a gap is not on the invoice: it is an outage nobody was told about. So the real question is what you must be alerted about, not what you can see on a dashboard.
Whatever replaces it has to be fed data by something, and that collection layer is the work. Plan to run the replacement beside the current system until the alerts have been exercised, not just configured.
Samples are made up and are not Datadog's prices. Type in the figures from your own invoice or order form.
The honest yearly cost of running open-source software yourself, set against the bill it replaces, with the seat count where it stops paying.
Paid seats against seats in use: what the idle ones cost a year, and what you save by right-sizing at renewal.
Arithmetic only, not tax, accounting or legal advice. Check the figures against your own order form.
Do not turn off the old system until the new one has caught a real or simulated failure. In outline:
List every monitor, dashboard and alert route. Mark the ones that notify a person and the ones that only draw a chart.
Compare the host count on the latest invoice with what is actually running. Remove agents from retired servers first; that alone changes the bill.
Decide how metrics will be collected and where they are stored, and who owns that system. Dashboards are the visible part; the collection layer is where the work is.
Stand up the new system on one service. Recreate that service's alerts and dashboards, and compare what each shows during normal traffic.
Trigger each critical alert on purpose in the new system, by stopping a process in a test environment, and confirm the right person is notified through the right channel.
Run both systems for at least a full month, including a deploy and a busy period, and compare incidents each saw.
Move the remaining services in small groups, updating the on-call runbook as each one moves.
Export dashboards and monitor definitions as files, confirm retention needs are met, then end the contract inside its notice window.
Run the vendor report for QuickBooks Online or Xero, then open the free audit.
General information, not legal advice. See cancelling without penalties. Last reviewed Oct 25, 2026. Product names belong to their owners and are used only to identify them.
Send us the export and see which lines are worth a second look. A connected version is planned. Datadog is one of the lines worth a look.
Worked example. Type over it.
Result
| Self-hosting, per year | Amount |
|---|---|
| Hosting, backups, tooling | 720.00 |
| Admin and restore-test time | 2,640.00 |
| Total | 3,360.00 |
| One-off set-up (not in the yearly total) | 1,200.00 |
The saving holds only if the time is real. Price the admin hours at what that person's time costs the company, not at zero.
Worked example. Type over it.
Result
“In use” is your call: last sign-in in the past 30 or 90 days from the vendor's admin report. A seat for someone on leave is not idle.