Dynatrace does not sell plans. It publishes a consumption rate card and your bill is whatever your usage multiplies it into. Hosts run $7 a month for Foundation and Discovery, $29 for Infrastructure Monitoring, and $58 per 8 GiB host for Full-Stack Monitoring. Logs and traces ingest at $0.20 per GiB. Synthetic monitoring is $1.00 per 1,000 HTTP requests or $4.50 per 1,000 browser actions, and real user monitoring is $2.25 per 1,000 sessions.
That is more transparent than most enterprise observability vendors manage, and it is still hard to turn into a number before you sign. The rate card has about a dozen meters running at once, and two of them behave in ways that catch people out badly enough to be worth the arithmetic below.
The Dynatrace rate card, meter by meter
These are the US dollar list prices published on Dynatrace's own pricing page, checked in August 2026. Volume commitments discount them, sometimes heavily, so treat list as your ceiling rather than your quote.
| Capability | List price | Billed as |
|---|---|---|
| Foundation & Discovery | $7/mo per host | $0.01 per hour per host |
| Infrastructure Monitoring | $29/mo per host | $0.04 per hour per host |
| Full-Stack Monitoring | $58/mo per 8 GiB host | $0.01 per memory-GiB-hour |
| Kubernetes Platform Monitoring | $1.40/mo per pod | $0.002 per hour per pod |
| Code Monitoring | $3.60/mo per container | $0.005 per hour per container |
| Log ingest | $0.20 per GiB | Volume ingested |
| Log retention | $0.0007 per GiB-day | Pay-per-query mode |
| Trace and event ingest | $0.20 per GiB | Volume ingested |
| Metric ingest | $0.15 per 100k datapoints | Datapoints |
| Real user monitoring | $2.25 per 1,000 sessions | Sessions |
| RUM with session replay | $4.50 per 1,000 sessions | Sessions with replay capture |
| Synthetic HTTP monitor | $1.00 per 1,000 requests | Requests, per location |
| Synthetic browser monitor | $4.50 per 1,000 actions | Actions, not runs |
| Runtime Vulnerability Analytics | $13/mo per 8 GiB host | Memory-GiB-hour |
| Security Posture Management | $5/mo per host | Per host |
Full-Stack Monitoring is priced on RAM, not on host count
This is the first thing to internalize, because the headline reads like a per-host price and it is not one. Full-Stack Monitoring is billed at $0.01 per memory-GiB-hour. The "$58 per 8 GiB host" figure is that rate worked through for a month: 8 GiB times $0.01 times 730 hours is $58.40. Double the memory and you double the bill.
| Host memory | Full-Stack per month | Infrastructure Monitoring per month | Cheaper option |
|---|---|---|---|
| 4 GiB | $29.20 | $29.20 | Break-even, exactly |
| 8 GiB | $58.40 | $29.20 | Infrastructure, by 2x |
| 16 GiB | $116.80 | $29.20 | Infrastructure, by 4x |
| 32 GiB | $233.60 | $29.20 | Infrastructure, by 8x |
| 64 GiB | $467.20 | $29.20 | Infrastructure, by 16x |
Infrastructure Monitoring is flat per host at $0.04 an hour, whatever the memory. So the crossover sits at exactly 4 GiB: below that, Full-Stack is the cheaper SKU, and above it Full-Stack grows without limit while Infrastructure stays put. On a fleet of 32 GiB database servers you are paying eight times more per host for Full-Stack than for Infrastructure.
That is not a criticism of the pricing, which is at least honest about what it charges for. It does mean the sensible pattern is to put Full-Stack only on the hosts where you genuinely need code-level tracing, and Infrastructure or Foundation on everything else. Teams that roll Full-Stack out fleet-wide by default are the ones who get a renewal shock.
Synthetic monitoring: the meter multiplies by interval and by location
Dynatrace is unusually precise in its documentation about how synthetic consumption is counted: one synthetic request is consumed each time a monitor runs, for each HTTP request executed, at each location. All three of those multiply.
An average month is 30.44 days, or 43,833 minutes, so one monitor running every minute from one location is 43,833 executions. At $0.001 per request that is $43.83 a month for a single check. Run it from three locations and it is $131.50. Add a redirect the monitor has to follow and every figure doubles again, because that is a second HTTP request per execution.
| What you are watching | Requests per month | Dynatrace cost |
|---|---|---|
| 1 monitor, every 5 min, 1 location | 8,767 | $8.77 |
| 1 monitor, every minute, 1 location | 43,833 | $43.83 |
| 1 monitor, every minute, 3 locations | 131,499 | $131.50 |
| 10 monitors, every minute, 3 locations | 1,314,990 | $1,314.99 |
| 50 monitors, every 5 min, 3 locations | 1,315,050 | $1,315.05 |
Browser monitors use a different unit again, and it is the one that surprises people most. Dynatrace charges $4.50 per 1,000 synthetic actions, and every click, navigation and assertion in a script is an action. A twelve-step checkout journey bills twelve units per run, not one. That makes shallow page-load checks cheap on Dynatrace and deep user journeys expensive, which is the opposite of how per-run vendors behave. We worked the crossover out in detail on our Dynatrace vs Splunk comparison, the AppDynamics vs Dynatrace comparison, where this same RAM rate is what makes Dynatrace expensive on database servers, and the Datadog vs Dynatrace comparison: the advantage flips to Datadog at about three actions per script.
Three worked bills
Rate cards are abstract, so here are three realistic shapes. All use list prices and ignore negotiated discounts.
A small platform team
Five application hosts at 8 GiB on Full-Stack is $292 a month. Add 50 GiB of logs at $0.20 for $10, and two HTTP monitors on a five-minute schedule from one location for $17.54. That lands near $320 a month, and the host meter is 91% of it.
A mid-size SaaS
Twenty hosts at 16 GiB on Full-Stack is $2,336. Add 500 GiB of log ingest at $100, 300,000 RUM sessions at $675, and twelve one-minute HTTP monitors from three locations at $1,578. That is roughly $4,690 a month before any security modules, and the synthetic line is now bigger than the log line.
Someone who only wants uptime
Ten public endpoints, checked every minute from three regions, with no APM and no logs. On the Dynatrace meter that is 1,314,990 requests, or $1,314.99 a month, for checks that still cannot run faster than once a minute. The same ten endpoints on a flat uptime plan cost $59 a month at 30-second intervals. If uptime is genuinely all you need, this is the comparison that matters, and it is not close.
What Dynatrace is worth paying for
None of the above is an argument that Dynatrace is overpriced. It is an argument that it is priced for a specific job. Automatic topology discovery, Davis AI root cause analysis and code-level tracing across a large mixed estate are things that would otherwise cost you headcount, and no uptime monitor replaces them. If you are trying to work out why a checkout call got slow across nine services, that is exactly what you are buying and it is worth the money.
The mistake is buying it to answer "is the site up", which is a question a dedicated external checker answers faster and for two orders of magnitude less. The two tools are complementary rather than competing, and there is a real argument for keeping the thing that tells you the site is down separate from the platform that might be down with it. That distinction is the subject of synthetic monitoring vs uptime monitoring, which is worth reading before you decide which budget this comes out of.
How to keep the bill predictable
Four things do most of the work. Put Full-Stack only on hosts that need code-level tracing and Infrastructure on the rest, because the 4 GiB crossover is unforgiving. Sample or filter logs at source rather than ingesting everything at $0.20 a GiB. Count your synthetic locations deliberately, since each one is a full multiplier on the meter. And keep an eye on the monthly consumption view rather than discovering the shape of your usage at renewal, the same discipline that makes the rest of a company's cloud and SaaS spend predictable instead of surprising.
The one thing you cannot buy your way around is the interval. Dynatrace HTTP monitors run every 1, 2, 5, 10, 15, 30 or 60 minutes, and one minute is the floor at any price. Datadog and New Relic have the same floor. If detecting an outage inside 30 seconds matters to your SLA, that is a product limit rather than a budget question, and it needs a different tool alongside.