Slowpoke vs Datadog

Datadog shows you everything happening in your system right now, and charges you for every host it watches. Slowpoke shows you one ordered list of what is slow, priced in seconds of waiting per day, on hardware you already own. Plenty of teams keep both.

The short answer

Which one you should pick

These two do different jobs, and most teams need both kinds of answer. If you have to pick one this week, here is how to tell which.

Choose Datadog if

You need to know what is happening right now

  • You carry a pager, and you need alerting, SLOs, incident management and an on-call rotation in one place.
  • You want logs, traces, real user monitoring, synthetics and security signals correlated for you.
  • You have budget approved, and a team big enough that a per-host bill costs less than assembling the same thing.
  • Sending production telemetry to a vendor's cloud is fine with your legal and security people.

Choose Slowpoke if

You need to know what to fix first

  • You have just taken over, and nobody can tell you what is actually running on the servers.
  • You need a number you can say out loud to a CEO, like 4,911 s of customer waiting a day. A p95 chart will not do.
  • You have no budget approved yet, and you need something useful before you ask for one.
  • Your production data has to stay on your own machines.
  • You need the work to actually happen, as a mission that only closes once the number stays down.

Side by side

Slowpoke and Datadog, line by line

Written to be fair, including where Datadog is the better tool. If a row here is wrong or out of date, that is a bug: open an issue and it gets corrected.

Aspect SlowpokeDatadog
What it answersWhat it costs you, and what to fix firstWhat is happening right now, across everything
Price€0, at any number of hostsFrom about $31 per host per month on an annual commitment (APM, 2026)
How the bill growsIt does notWith hosts, ingested spans and retention
Host countingThere is noneMeasured hourly, top 1% of hours discarded, month billed at that peak
Where your data livesYour machines, alwaysDatadog's cloud
Time to first answerAbout 15 minutes, one command per machineAn agent per host, then the dashboards you build
Prioritised for youOne list, in seconds of waiting per dayDashboards, percentiles and monitors you define
Query plans with a fixReal EXPLAIN, twelve named diagnosesDatabase Monitoring shows plans, without a recommendation
Work that verifies itselfMissions closed by measurementNo
Report for non-engineersEffort, return per hour, budget, DORABuilt by you, by hand
Live incident responseNot its jobAlerting, on-call, SLOs, incident management
Logs, RUM, session replay, syntheticsNeverAll of it
RetentionAs long as your disk allowsWhat your plan includes

The bill grows with the problem you are trying to solve

Datadog APM is billed per host: around $31 per host per month on an annual commitment as of 2026. The host count is not an average either. Datadog measures it every hour, throws away the top 1% of hours, and bills the whole month at what is left of the peak. Autoscaling to survive a traffic spike is therefore also a billing event.

The uncomfortable part is the direction it points. The traffic that makes your application slow is the same traffic that raises your bill, so the more urgent the performance problem, the more expensive it becomes to look at it. Slowpoke is a Go binary on hardware you already pay for, so adding hosts adds nothing to what it costs you.

Datadog tells you everything. Slowpoke tells you what to do.

Open an APM page in Datadog and you get p50, p75, p95, p99, throughput, error rate, a flame graph and a service map. Every one of those is true, and not one of them tells you which of your two hundred endpoints deserves Thursday afternoon.

Slowpoke answers only that question. It prices every endpoint, query, queued job and scheduled command in the same unit, seconds of waiting per day, and sorts them. A query that takes 40 ms and runs 90,000 times a day costs 3,600 seconds of waiting per day: more than a 30-second report that runs once. On a percentile dashboard the report looks like the emergency. In Slowpoke it is somewhere around row twenty.

The same unit works on people who do not read query plans. "The checkout is at p95 1.9 seconds" usually starts an argument. "Checkout costs our customers 412 seconds of waiting every day, and 42 minutes of work removes it" is something a CEO can say yes to.

The first day on a new job

Datadog assumes somebody already knows the system. You install the agent on the hosts you know about, and you build the dashboards you already know you want.

Slowpoke assumes the opposite, because it is built for the engineer who just arrived. slowpoke-agent detect reads a machine and tells you what is on it: the nginx server blocks and whether their log format records how long a request took, the PHP-FPM pools, the local MySQL and PostgreSQL, Redis, and the applications with their framework. It only reads files, so it is safe to run on production on your first morning.

Then it tells you what it cannot see, which a dashboard never does. A database nobody created a user for, a pool with its slow log switched off, a service with no framework package: each one is listed with the exact lines that close the gap, and the lines that undo them.

Can you run both?

Yes, and it is the common case. Keep Datadog for the pager, the logs and the 3 a.m. question of what just broke. Add Slowpoke for the quarterly question of where the time goes and what the team should fix. It costs nothing to run alongside and never touches your request path. Its debt report is the document you take to the budget meeting where you defend the Datadog invoice.

FAQ

Slowpoke and Datadog, asked plainly

Is Slowpoke a Datadog alternative?
For the part of Datadog you use to find slow endpoints and slow queries, yes, and it is free and runs on your own machines. For alerting, on-call, logs, real user monitoring, synthetics and security, no, and it never will be. Slowpoke is deliberately narrow.
How much does Slowpoke cost next to Datadog?
Nothing. There is no hosted plan, no per-host price, no seat count and no event count. You pay for the machines you already run. A Datadog APM bill starts at roughly $31 per host per month on an annual commitment and grows with ingested spans and retention.
Can Slowpoke replace Datadog entirely?
Only if the parts you actually use are the APM and database views. If your team is paged through Datadog, ships logs to it, or relies on its RUM and synthetics, replacing it with Slowpoke would leave real holes. Many teams shrink their Datadog plan rather than cancel it.
Does anything leave my infrastructure?
Only traffic to your own Slowpoke server, which you host. Query values are redacted before they are sent, and the real statements used for EXPLAIN never leave the machine that saw them. The agent never runs SQL chosen by the server.

Find out what you inherited

Give it fifteen minutes on one server and the first number is yours. Finding out costs nothing.