Experimenting with Orchestration¶
The GP4L team has adopted the definition of Orchestration as means to transform end-to-end processes vs Automation that is focusing on the streamlining of individual tasks inside processes.
This means that orchestration involves the design and implementation of an intelligent, multi-step process or a set of (sub)processes that work together to accomplish a higher-level goal.
The design of a flexible process that clearly defines the tasks that are implemented in each step together with the interaction with other components to get or push information to is essential for high-performing, reliable digital transformation of the traditional manual processed defined via standard operational procedures.
To achieve this in addition to the process design, one must choose a suitable orchestrator that will drive the processes, follow its execution and ensure overall status consistency at all times.
While working on different use-cases the GP4L team has been able to investigate the capabilities of the following orchestrators:
Camunda¶
Camunda is a universal orchestrator that enables definition of complex process flows that are defined using Business Process Model Notation (BPMN). Each process tasks implementation can be done in Java, Go, Node.js, Python or C#. It offers a GUI that provides the general information about all processes including status and tracking information as well as statistics regarding the performances and use of each process.
Camunda started as a fork of the popular Activiti solution and is available as free open-source solution when self-managed. The core of the Camunda Platform is source-available, while additional tools are free for non-production use.
To see how GP4L has used Camunda in action take a look at the following labs:
AirFlow¶
Apache AirFlow is a community based platform for developing, running and monitoring workflows. It is a scalable solution that enables dynamic instantiation of workflows defined in Python. The accompanying UI provides a detailed information regarding the state of each workflow including logs. Using operators one can integrate different services into the workflow and use various programming languages to define different tasks of the workflow.
The AirFlow community is quite active and provides detailed documentation on the large number of features. Thus, new versions of AirFlow are released frequently.
To see how GP4L has used AirFlow in action take a look at the following labs: