Dapr in Practice - Marc Klefter - NDC Oslo 2024

Dapr is a Distributed Application Runtime that provides building blocks for cloud-native applications, with most deployments (90%+) running on Kubernetes

Key building blocks include:

• Service invocation
• Pub/Sub messaging
• State management
• Workflows
• Bindings for external system integration

Applications interface with Dapr through a sidecar architecture, providing location transparency and decoupling from infrastructure details

Components are pluggable implementations that enable Dapr building blocks to work with specific technologies (e.g., Redis for state, SNS/SQS for messaging)

Dapr is not an all-or-nothing proposition - teams can adopt building blocks incrementally as needed

While Dapr provides infrastructure abstraction, teams still need to handle:

• Service naming and discovery
• Component provisioning and configuration
• Access control policies
• Security considerations

The workflow feature enables modeling of complex business processes with built-in durability and error handling

Dapr can work alongside service meshes, though there is some overlap in capabilities

Key benefits include:

• Portability between environments/clouds
• Standardized building blocks across languages
• Separation of business logic from infrastructure code
• Built-in observability and security features

Common adoption pattern is to start with a modular monolith using messaging, then gradually extract services leveraging Dapr building blocks