Obseria vs Dynatrace

Feature comparison

Obseria as a Dynatrace alternative: the short answer

Dynatrace and Obseria are architecturally closer than most platforms in this market. Both store telemetry in an index-free lakehouse with schema-on-read, both use a pipe-based query language, both build a live topology automatically and both apply causal AI on top of it. If you are evaluating a Dynatrace alternative, the decision therefore rarely hinges on raw capability. It hinges on the commercial model, on whether you want proprietary auto-instrumentation or OpenTelemetry at the core, and on where your data has to live. The facts below are based on Dynatrace's public pricing page and documentation as of October 2026.

Pricing model: platform subscription vs. pay per signal

Dynatrace sells the Dynatrace Platform Subscription (DPS). According to its pricing page, customers commit to a minimum annual spend and draw it down by consuming any capability at rate-card prices. Larger commitments lower the unit price, usage beyond the commitment is billed at the same rates, and seats are unlimited. The rate card uses several units: hosts or host-hours for infrastructure monitoring, memory-GiB-hours for full-stack monitoring, pod-hours for Kubernetes, sessions for real user monitoring, and GiB ingested, GiB-days retained and GiB scanned for logs, traces and events. For logs you can choose between pay-per-query and retention with bundled queries.

The strength of DPS is flexibility inside a single budget: teams can shift spend between capabilities without renegotiating. The trade-off is that you plan an annual commitment up front, and forecasting requires modeling memory sizes, pod counts, data volumes and query behavior.

Obseria's model is purely usage-based and has fewer dimensions:

• Logs, spans and RUM web events: $0.060 per million ingested, $0.540 per million stored, 30-day retention.
• Metric data points: $0.020 per million ingested, $0.180 per million stored, 13-month retention.
• Synthetic API check runs: $0.20 per thousand. Intelligence agent credits: $0.60 each.
• No host, memory or seat dimension. Filtered data costs ingestion only, and every feature is included.

Organizations with a central procurement process and stable growth often like a committed platform budget. Teams that want to start small, pay only for actual data and avoid sizing hosts by memory often prefer per-signal pricing.

Data model and storage

Dynatrace Grail is described by Dynatrace as a data lakehouse that stores data without indexes or schema management, queried with the Dynatrace Query Language (DQL). Obseria Lake follows the same principle: index-free, schema-on-read, petabyte scale, with field-level privacy controls. In practice both avoid the index planning that older log platforms require.

Differences show up at the edges. Obseria offers three query languages: OQL for pipe-style exploration across all signals, PromQL for teams coming from Prometheus, and SQL for analysts. Obseria Pipeline filters, enriches, masks PII and routes data at ingest, and filtered data is billed for ingestion only, which makes it straightforward to keep noisy data out of storage.

fetch logs
| filter service == "payments"
| summarize count(), by:{level}

Instrumentation, OpenTelemetry and lock-in

Dynatrace is known for OneAgent, which instruments hosts, processes and many runtimes automatically with little configuration. That depth of automatic instrumentation is a real strength, especially in large estates with legacy applications. Dynatrace also ingests OpenTelemetry data, so teams can combine both approaches.

Obseria puts OpenTelemetry at the center. OTLP is the native protocol, semantic conventions are the data model, and you can export your data via OTLP at any time. Teams with existing OpenTelemetry SDKs usually see their first trace within 15 minutes. The Obseria Agent is installed once per host for metrics, logs, traces, profiles and security signals, and continuous profiling runs at under 1% overhead. Tail-based sampling keeps the traces that matter while controlling volume.

The lock-in consideration is simple: instrumentation built on OpenTelemetry is portable between backends; instrumentation that depends on a vendor agent is not portable in the same way.

AI and root-cause analysis

Dynatrace has used causal AI in observability for many years with Davis, which uses the Smartscape topology to determine root causes. Its current pricing page also lists agentic operations billed in AI units, with different unit costs for chat, query translation and autonomous investigations.

Obseria Intelligence combines causal, predictive and generative AI on top of Obseria TopoGraph. Agentic root-cause investigations return a chain of evidence that engineers can verify step by step. Agent credits cost $0.60 each, and customer data is never used to train AI models. If causal AI is already central to your operations, both platforms give you that model; compare them on your own incidents during a trial.

Data residency and compliance

Dynatrace SaaS runs on AWS, Microsoft Azure and Google Cloud, with European regions available; according to its data security documentation some regions, including AWS Frankfurt, are available on request. Dynatrace also offers a Managed deployment model for customers who need to operate the platform themselves.

Obseria stores data in the EU (Frankfurt, Dublin), in the UK (London) or in your own cloud account (BYOC). Certifications include ISO 27001, 27017 and 27018, SOC 2 Type II, BSI C5 for the EU region and TISAX, with support for DORA, NIS2 and GDPR, automatic PII masking, field-level access, audit logs and a 99.95% SLA.

Migrating from Dynatrace to Obseria

Because OneAgent instruments automatically, the main migration task is replacing that coverage. Plan it in stages: deploy the Obseria Agent or OpenTelemetry Collectors next to OneAgent on a pilot cluster, enable OpenTelemetry auto-instrumentation for your main runtimes, and translate the DQL queries, dashboards and alerts your teams rely on into OQL. Run both in parallel until alert parity is confirmed, then expand cluster by cluster.

For cloud services, Obseria monitors more than 300 AWS, Azure and GCP services agentlessly, and the Integrations Hub covers 700+ technologies, so much of the infrastructure coverage does not depend on a host agent at all. Use the 14-day free trial, which requires no credit card, to validate coverage and alert behavior on real incidents before you commit.

Which tool fits you?

Switching in three steps

1. Pilot next to OneAgent. Deploy the Obseria Agent or OpenTelemetry Collectors on one cluster alongside OneAgent and compare coverage.
2. Translate DQL to OQL. Both are pipe-based, so most queries, dashboards and alert conditions map closely; start with on-call views.
3. Expand by cluster. Once alert parity is confirmed, roll out cluster by cluster and retire OneAgent where OpenTelemetry covers the workload.