Flow supports multiple backend schedulers which configure how work is scheduled in the workflow.
By default, flows are configured to use an active scheduler which runs as a process. The active scheduler:
Determines the order that layers should be executed in.
Submits layers to the
Waits on layers to complete to submit new layers for execution.
it actively manages the tasks in the workflow until their completion and then exits.
from laminar import Flow from laminar.configurations import schedulers class ScheduledFlow(Flow): ... flow = ScheduledFlow(scheduler=schedulers.Scheduler())
Delegated schedulers are experimental.
Delegated schedulers don’t manage the execution of a workflow. Instead, they compile the flow’s directed acyclic graph into a format that can be used by other schedulers to manage the workflow.
Delegated schedulers provide a method for writing flows easily in
laminar and executing them on another workflow engine.
AWS Step Functions¶