Saga and Workflow Runtime
Define linear sagas or branching node graphs, execute them through a simple HTTP API, and let Trama handle retries, compensation, async callbacks, and execution state.
Trama coordinates distributed workflows while keeping services focused on business logic. No SDKs. No framework integration. No platform migration required.
Comparison
Distributed workflows become hard to reason about when orchestration logic lives inside individual services.
Trama sits between hand-rolled orchestration and full workflow platforms, without requiring a platform adoption project.
Trama injects a callback URL into async requests and resumes execution when the callback is received.
Execution Walkthrough
A stored workflow definition is created, execution starts, a switch node evaluates the payload and picks a branch, the selected branch fires an async task, execution pauses waiting for a callback, the callback resumes it, and the workflow completes.
The same execution model works with any service that can receive HTTP requests.
Event log
Trama evaluates the switch node, routes to the pix branch, pauses on the async step, validates the callback token, resumes execution from the next node, and stores the completed state through the API.
Architecture
Task nodes call services. Switch nodes choose execution paths. Async nodes pause and resume through callbacks. Sleep nodes pause execution for a set duration.
Trama separates orchestration from execution. The API stores execution state and enqueues work before handing it to workers. Workers dispatch each node to the appropriate handler — task nodes make HTTP calls, switch nodes evaluate JSON Logic expressions and route to the next node.
Async task nodes pause execution and wait. When the external service calls back with a signed token, the callback receiver validates and re-enqueues the execution to resume.
Trama integrates through HTTP and does not require changes to existing services.
Capabilities
Trama provides the primitives required to safely coordinate distributed workflows in production.
Execute both sequential flows and conditional workflows using switch nodes.
Pause execution on async steps and resume via signed callbacks (or timeout).
Built-in retry and compensation handling across all workflow steps.
Inject payload and step results into HTTP requests using templates.
"orderId": "{{payload.orderId}}""reservationId": "{{nodes.reserve.response.body.id}}"Persist execution state and inspect workflow progress through the API.
Expose metrics and traces for workflow execution, failures, and callbacks.
Trama exposes an HTTP API and OpenAPI contract. Workflows can be triggered and inspected from any language or platform. No SDK, framework integration, or platform migration is required.
Inspect executions, debug step-by-step with Gantt timelines, and create or edit definitions with a built-in graph editor. Explore the UI
Comparison
There are three common approaches to distributed workflows.
| Approach | How it works | Tradeoff |
|---|---|---|
| Hand-rolled orchestration | Workflow logic spread across services with inconsistent retry and rollback behavior. | Fast to start, difficult to maintain as workflows grow. |
| Trama | Dedicated workflow runtime. Linear sagas, branching node graphs, and async task coordination through a simple HTTP API. | Centralizes workflow coordination without turning it into a platform project. |
| Heavy workflow engines | Powerful orchestration platforms designed for complex workflow management and platform-level automation. | Broader capabilities, but more operational and adoption overhead. |
Use Cases
