What is Dapr?
Dapr is a distributed application runtime that provides durable, verifiable execution for workflows and AI agents, so applications survive crashes, restarts, and outages and resume exactly where they left off. Dapr is a graduated CNCF project built on infrastructure-agnostic APIs with production-grade security, resiliency, and observability, and it bundles durable workflows, agentic AI, pub/sub, state management, and other distributed-system building blocks into one set of integrated APIs.
Key Features
- Durable execution: Every step of an application's work is checkpointed as it runs, so when a process crashes, a pod restarts, or a network call fails, Dapr resumes from the last completed step instead of starting over or leaving work half-finished.
- Durable Workflows: Orchestrate long-running, stateful processes that always run to completion, with retries, timers, human approvals, and sagas, without operating a separate workflow engine. Supported patterns include task chaining, fan-out/fan-in, monitor, timers, and external system interaction, and workflow state can be stored in any database of choice.
- Durable AI Agents: Build AI agents that reason, call tools, and recover from any point of failure, because every step of an agent's plan is checkpointed and resumable.
- Microservices APIs: Service-to-service calls with built-in service discovery, mTLS, retries, and distributed tracing through a single API.
- Event-Driven Architecture: Pub/sub messaging that decouples publishers from subscribers and works across any supported message broker without code changes.
- Integrated building-block APIs: Out-of-the-box durable workflow, agentic AI, pub/sub, state management, secret stores, external configuration, bindings, actors, jobs, distributed lock, and cryptography, with built-in security, reliability, and observability so teams avoid writing boilerplate for production-ready applications.
- Developer tooling and learning resources: A Dapr Dev Dashboard for inspecting local Dapr workflow executions, plus Dapr University, a free and self-paced learning program.
Use Cases
- Business process automation: Author workflows in code to automate complex business processes that are stateful, durable, and long-running, combining Dapr workflow with Dapr APIs to send messages, invoke services, and store key/value data.
- AI agent development: Build agents that reason, call tools, and resume from the last completed step after any failure, rather than restarting an entire plan.
- Microservices teams: Replace hand-written service discovery, mTLS, retry, and tracing code with a single service-to-service invocation API.
- Event-driven systems: Connect publishers and subscribers through pub/sub messaging across supported message brokers without changing application code.
- High-volume and IoT workloads: Documented customer examples include tracking billions of messages on IoT devices with Kubernetes, handling millions of transactions efficiently, near-real-time personalized recommendations at scale, and improved vulnerability scanning.
Who It Is For
Dapr is aimed at developers and platform teams building distributed applications, microservices, long-running business workflows, and AI agents, particularly those running on Kubernetes or across mixed infrastructure. It solves the problem of writing and maintaining boilerplate for resiliency, security, observability, and state handling, and of losing in-progress work when processes crash or restart. The source cites a 30% increase in developer productivity from its out-of-the-box features.
Pricing
The source material does not state Dapr's pricing model. Dapr is described as a graduated CNCF project, and Dapr University is described as a free and self-paced learning program; the site also notes that Dapr support is provided by Diagrid.
Frequently Asked Questions
What is Dapr used for?
Dapr is used to build distributed applications with durable execution. Developers use Dapr for durable workflows, durable AI agents, microservice service-to-service calls, and event-driven pub/sub messaging, along with state management, secret stores, bindings, actors, jobs, distributed lock, and cryptography APIs.
How does Dapr durable execution work?
Dapr checkpoints every step of an application's work as it runs. When a process crashes, a pod restarts, or a network call fails, Dapr resumes from the last completed step instead of starting over or leaving work half-finished.
Is Dapr free?
The source material does not state Dapr's pricing. Dapr is a graduated CNCF project, and Dapr University is offered as a free, self-paced learning program.
What can developers build with Dapr?
Dapr supports durable workflows for long-running stateful processes, durable AI agents that recover from any point of failure, microservices with built-in service discovery, mTLS, retries, and distributed tracing, and event-driven architectures with broker-agnostic pub/sub.
Who should use Dapr?
Dapr suits developers and platform teams building microservices, long-running business workflows, or AI agents who want built-in security, resiliency, and observability without writing boilerplate. It is particularly relevant for teams running distributed applications on Kubernetes or across infrastructure-agnostic environments.