RousselTMLEARNING

Observability

Dynatrace: definitions, architecture and Davis artificial intelligence

The basic concepts of Dynatrace (entities, Smartscape, PurePath, Davis) and its architecture: OneAgent, ActiveGate, SaaS or Managed cluster, and the role of the Davis AI in detecting problems.

1. Definitions to know #

Dynatrace vocabulary.

Glossary #

TermDefinition
Environment (tenant)Isolated space that receives and stores the data of a scope (for example production)
EntityAny detected object: host, process, process group, service, application, container...
OneAgentSingle agent installed on the host: it automatically discovers and instruments processes and services
ActiveGateRelay and gateway: routing agent traffic, cloud and Kubernetes monitoring, extensions, private synthetics
SmartscapeMap of dependencies between all entities, built in real time
PurePathEnd-to-end distributed trace of a request, with code-level detail
DavisAI engine that detects anomalies, groups events and identifies the root cause
GrailData lakehouse that stores logs, metrics, traces and events, queryable in DQL

2. The architecture #

Agents, gateways, cluster.

Diagram #

Dynatrace architecture
OneAgents can send directly to the cluster or go through an ActiveGate.
  • The OneAgent is installed once per host: it detects technologies (Java, .NET, Node.js, PHP, Go, web servers, databases) and injects instrumentation without code changes.
  • The ActiveGate concentrates network flows (useful behind a firewall), queries cloud and Kubernetes APIs and runs extensions.
  • The cluster stores and analyzes: hosted by Dynatrace (SaaS) or on your premises (Managed).
  • Data can also arrive through OpenTelemetry, Prometheus, the API or extensions.
Good to know: The current platform (Grail, apps, AppEngine) coexists with the « classic » interface. Both approaches are covered in the course.

3. Davis, the artificial intelligence #

From thousands of events to one problem.

Principle #

Davis automatically learns the normal behavior of each entity (baseline) and detects deviations: rising error rate, slowdowns, saturation. Thanks to Smartscape, it links anomalies together and presents a single problem with its root cause and its impact on users.

AIRole
Causal AITopology analysis to trace back to the root cause
Predictive AIForecasts (disk saturation, capacity)
Generative AI (Davis CoPilot)Help writing DQL queries and with analysis
Good practice: Davis is all the more relevant when the topology is complete: install OneAgent in Full-Stack mode on all hosts of an application rather than on just a few.

Related articles

Accessing the Dynatrace playground and getting started

The Dynatrace playground is a demo environment full of real data: how to access it, what it allows and does n…

Grail buckets in Dynatrace: creation, routing, retention, rights and costs

Complete guide to Grail buckets: purpose, default buckets, creation in Storage management, with the API or Te…

The Dynatrace QuickStart app: getting your first data in minutes

Discover the QuickStart app: add data with Auto Discovery or a dedicated tile, read Findings, explore with da…

Keep learning

Support us

♥ SUPPORT US

Support RousselTM Learning

Courses, articles, labs and certificates stay free with no intrusive ads. A voluntary contribution, of any amount you choose, pays for hosting, lab environments and new content.

Contribute with PayPal Any amount · card or PayPal accountLearn more
QR code PayPal
Or scan this QR code with your phone

Thank you, every contribution counts!