Dynatrace

What Dynatrace actually is

Dynatrace is an enterprise observability platform built around one idea: the monitoring system should find the problem itself. Founded in 2005 in Linz, Austria by Bernd Greifeneder and Sok-Kheng Taing, headquartered in Waltham, Massachusetts, and listed on the NYSE as DT since its August 2019 IPO, the company now reports over $1.7 billion in annual recurring revenue and more than 5,200 employees. That scale is part of the product: this is a platform with its own causal-AI engine (Davis), its own massively parallel telemetry store (Grail), and a single agent (OneAgent) that discovers and instruments whatever it lands on — hosts, processes, services, Kubernetes workloads, serverless functions — with minimal manual configuration.

The workflow is the differentiator. OneAgent deploys and starts collecting metrics, logs, traces and topology automatically; Smartscape draws the dependency map; Davis analyzes the whole stack and, when something breaks, surfaces a root cause rather than a wall of correlated alerts. You spend less time wiring alert thresholds and more time reading the answer. Davis CoPilot, the generative layer that reached general availability in October 2024, adds natural-language queries and assisted investigations on top of that deterministic engine.

Coverage spans application performance monitoring with code-level profiling and distributed tracing, infrastructure monitoring, real user and synthetic monitoring, log management and analytics, application security (runtime vulnerability analytics and runtime application protection), security posture management, business analytics, and database and mainframe monitoring. The platform claims 900+ supported technologies and integrates with the major clouds and OpenTelemetry pipelines. We file it under monitoring hosting; for the hosts you would point it at, see our benchmarks.

Pricing: what you really pay

Dynatrace pricing is honest in structure and demanding in magnitude. Everything runs through one construct — the Dynatrace Platform Subscription (DPS): a minimum annual spend commitment at the platform level, drawn down against a published rate card, with no per-seat fees, no product tiers gating features, and no penalty for exceeding the commitment (overage bills at the same rate-card price). Negotiated discounts scale with commitment size, and real contracts are typically one to three years.

The published list rates, straight from dynatrace.com/pricing as we write: Foundation & Discovery at $7/mo per host ($0.01 per host-hour) for basic infrastructure health and inventory; Infrastructure Monitoring at $29/mo per host ($0.04 per host-hour); Full-Stack Monitoring at $58/mo per 8 GiB host ($0.01 per memory-GiB-hour) — that is the tier with APM, automated root-cause analysis for end-to-end transactions and code-level profiling. Kubernetes Platform Monitoring is $1.40/mo per pod (included on Full-Stack hosts), Code Monitoring $3.60/mo per container, log ingest $0.20 per GiB with retention at either $0.0007 per GiB-day plus $0.0035 per GiB scanned (pay-per-query) or $0.02 per GiB-day with queries included, metrics ingest $0.15 per 100,000 data points, traces and events $0.20 per GiB to ingest. Real User Monitoring runs $2.25 per 1,000 sessions (with Session Replay at $4.50), synthetic HTTP monitors $1.00 per 1,000 requests and browser monitors $4.50 per 1,000 actions. Security adds $13/mo per 8 GiB host for runtime vulnerability analytics and $5/mo per host for security posture management.

The catch is the meter. Full-Stack billing follows host memory with a 4 GiB minimum and rounding to the next 0.25 GiB, measured in 15-minute intervals: an 8 GiB host is about $58/mo, a 16 GiB host about $117/mo, a 32 GiB host about $234/mo. A worked example for a 20-app-host team with supporting infrastructure, 200 pods, 500 GiB of logs and 300,000 RUM sessions lands around $3,700/month at list — roughly $45,000 a year as the commitment input, before application security or session replay. Large enterprises negotiate 25–50% off list. There is no permanent free plan; a 15-day trial is the only unpaid path.

Our view: the rate card is published, the meters are documented, and the sales process is comparatively straight — that is more transparency than this category offered five years ago. But you are buying an enterprise platform, and the bill scales with the size of your machines, not just their count. Budget with real host memory, not the headline.

Performance & infrastructure

Dynatrace runs the platform as a service on its own telemetry backbone, so there is no infrastructure for you to size. What matters instead is analytical performance: how fast the system can ingest, correlate and answer. Grail, introduced in 2022, stores high-cardinality metrics, logs, traces and events together with topology context, which is what lets Davis trace a symptom across tiers instead of per-tool alerting. Retention is generous and adjustable — 15 months of metrics at one-minute granularity, log and trace retention controllable from days to 10 years.

The operational story is strong. OneAgent is genuinely low-effort: install it, and it discovers processes and frameworks, captures code-level detail where supported, and keeps instrumentation current through updates. For teams with hundreds of hosts and frequent deployments, that automation is the difference between a monitoring stack and a monitoring team. OpenTelemetry ingest is supported for teams standardizing on OTLP, and the platform covers hybrid estates — cloud, Kubernetes, on-prem, mainframe — under one topology.

Compliance and governance are packaged rather than bolted on: SSO, granular access policies, IP allow lists, built-in data masking, configurable retention and hard deletion for GDPR and CCPA, with certifications including ISO 27001, SOC 2 Type II, FedRAMP Moderate and HIPAA available under DPS contracts. Unlimited seats are included at every commitment size.

Reliability of the vendor itself is not in question: two decades of operation, a public company with audited financials, and a large engineering base split across Linz, Gdansk and Barcelona. Our benchmark program covers infrastructure providers rather than observability suites, so we have no hands-on Dynatrace measurements of our own — we are describing documented capabilities and published rates, not test output.

Where it falls short

The main one is price. Even at list with no surprises, a modest production estate crosses five figures a year quickly, and the annual-commitment requirement means there is no month-to-month exit ramp. Startups and small teams are simply not the target; open source stacks or lighter SaaS tools make more sense until the cost of an incident exceeds the cost of the platform.

Second, capability sprawl. Full-Stack plus application security plus logs plus RUM plus synthetics is a lot of meters to watch, and while overage is not penalized, the drawdown model rewards forecasting skill. Teams that treat the rate card as a menu rather than a budget end up in uncomfortable renewal conversations. Memory-banded host pricing also punishes large-memory machines — a fleet of 32 GiB database hosts costs multiples of what the same host count does on flat per-host tools.

Third, depth comes with weight. The query language (DQL), the configuration surface and the sheer number of platform concepts take weeks to learn properly, and the auto-discovery approach means less explicit control over what gets instrumented and billed. Teams that want minimal, hand-tuned monitoring can find it more than they need.

Finally, everything routes through the vendor's cloud: there is no self-hosted deployment of the modern platform. Organizations with strict data-residency or air-gap requirements have to evaluate that against policy early.

Who should pick this

We pick Dynatrace when downtime is expensive and the estate is complex: mid-market and enterprise teams running distributed applications across Kubernetes and VMs who want automatic discovery, causal root-cause analysis, and one platform for APM, infra, logs, RUM and runtime security. It is our pick where an on-call rotation wastes hours correlating dashboards — the automation pays for itself in incident minutes saved. Regulated industries get real value from the bundled compliance posture and audit story.

We would not pick it for a two-person startup, a single static site, or a team that wants to hand-roll Prometheus and Grafana for a fraction of the cost and accept the operational overhead that comes with that choice. If your bill would land under a few hundred dollars a month of capability usage, you are buying machinery you will not drive.

Key takeaways

• Enterprise observability platform founded 2005 in Linz, Austria; NYSE-listed (DT) since 2019, $1.7B+ ARR, headquartered in Waltham, Massachusetts.
• Davis causal AI plus OneAgent auto-discovery do the correlation work: root causes and dependency maps instead of raw alert storms.
• DPS licensing: one annual platform commitment against a published rate card — Full-Stack $58/mo per 8 GiB host, Infrastructure $29/mo per host, Foundation & Discovery $7/mo per host, logs $0.20/GiB ingest.
• Billing follows host memory with a 4 GiB floor, so host sizing — not host count — drives cost; a 32 GiB host runs about $234/mo at list.
• No permanent free plan (15-day trial only), no month-to-month option; enterprise contracts are negotiated, typically 25–50% below list at scale.
• Best fit for complex distributed estates where incident cost justifies the platform; wrong for small teams and budgets.

Sources

• https://www.dynatrace.com/pricing
• https://www.dynatrace.com/careers/about
• https://ir.dynatrace.com/news-events/press-releases/detail/144/dynatrace-announces-pricing-of-initial-public-offering
• https://last9.io/blog/dynatrace-pricing
• https://usagepricing.com/blueprint/dynatrace
• https://monitoringcost.com/dynatrace-pricing