Processes

Processes coordinate tasks, decisions, and integrations across systems and people, with full visibility into every running instance. Aletyx models processes with the BPMN standard and executes them on its enterprise build of jBPM and Kogito, running cloud-natively on Quarkus.

​

BPMN modeling

BPMN (Business Process Model and Notation) is an OMG standard that represents workflows as graphical nodes and connectors. It defines three core categories of flow objects:

• Events — start, intermediate, and end events that begin, interrupt, or conclude a flow (including messages, signals, and timers).
• Activities — the work itself: tasks and subprocesses.
• Gateways — branching and merging logic for parallel execution and conditional flow.

BPMN is primarily a visual specification of expected behavior; the jBPM/Kogito engine supplies the runtime execution semantics. See the BPMN reference and Start with BPMN.

​

Task types

​

Common patterns

• Rules orchestration — sequence multiple decisions within a process.
• SAGA — coordinate long-running, distributed transactions with compensation.
• Ad-hoc subprocesses — model unstructured work without a fixed order.

​

Adaptive process architecture

Aletyx’s compact (adaptive) architecture intelligently groups related services into unified deployments — avoiding the choice between a rigid monolith and a fully distributed microservices sprawl. Its components:

• Process definition models — BPMN files compiled into optimized executable code.
• Orchestration engine — the jBPM engine reimagined for the cloud, coordinating parallel execution, conditional branching, and event-driven flow.
• Cloud-native runtime — built on Quarkus for fast startup, handling transactions, API exposure, security, and lifecycle.
• Data Intelligence Layer (Data-Index) — real-time process visibility via GraphQL without impacting operational performance.
• Process History Service (Data-Audit) — immutable audit trails for compliance, with replay and analysis.
• Temporal Event Coordinator (Jobs Service) — timers, deadlines, and SLA enforcement.
• Human Collaboration Framework — task assignment, permissions, and human-in-the-loop interactions.
• Persistent Context Store — a relational database holding consistent process state across components.

The result is near-linear scalability with far less operational overhead than full microservices, sub-millisecond response times for critical in-memory calls, and comprehensive audit trails — well suited to stateful, human-in-the-loop, and regulated workloads.

​

Event-driven processes

Aletyx ships out-of-the-box Kafka and CloudEvents integration, so processes can consume and emit events with minimal boilerplate:

1. A CloudEvent arrives matching a configured BPMN message node type.
2. Kogito routes it to the correct process instance and injects the event data into the process context.
3. The process executes its tasks.
4. At key checkpoints, the process can emit new events to trigger downstream services.

Event metadata such as type and source is embedded automatically for correct routing. See the event-driven guide.

​

Calling processes from code

By default Kogito generates REST endpoints for your processes, but you can drive them directly from Java when you need custom validation or want to call from other services — using CDI injection and the ProcessService API. See Invoking Kogito processes with Java APIs.

​

Deployment & monitoring

Deploy processes to Kubernetes, OpenShift, or Docker with multi-environment promotion. The Data-Index and Data-Audit services give real-time visibility into running instances and an immutable history of every step for compliance. See deployment options.

​

Next steps