Netflix Conductor, Temporal (Uber Cadence), Zeebe, and Airflow were evaluated for their use in various business scenarios, including data pipelines, AI/ML workflows, customer support workflows, and developer tooling purposes. Pre-requisites for the engine included supporting multiple languages, high scale of invocation, concurrency, and throughput, durability for long running workflows, operational excellence, small footprints, enterprise support, and an open source permissive license. General purpose orchestration engines were preferred, with specific engines like Airflow and Zeebe becoming non-options due to their limitations. BPMN was not a fit for the companys needs. AWS Step Functions was very tied to serverless applications and AWS related services. Support was also a priority for the company.
