{no title found}

A sub-resource of a job resource, the result of a successful run of that job.

An execution resources may be retrieved by any site; however, it may be created and updated only by the site that ran its job.

A job is a resource that isolates workflow configuration from execution.

Rather than directly requesting a workflow execution by supplying a workflow ID along with configuration, one creates a job that pairs a workflow with configuration. Then, a workflow is executed by supplying a job ID without additional configuration.

A job can have multiple executions, and a workflow's executions are precisely the executions of all jobs created for that workflow.

An operation is a resource for tracking a request to run a job.

When a request to run a job is issued, an operation resource is created.

An operation is akin to a "promise" or "future" in that it should eventually resolve to either a successful result, i.e. an execution resource, or to an error.

An operation is parameterized to return a result type, and a metadata type for storing progress information, that are both particular to the workflow type.

Operations may be paused, resumed, and/or cancelled.

Operations may expire, i.e. not be stored indefinitely. In this case, it is recommended that execution resources have longer lifetimes / not expire, so that information about successful results of operations are available.

A site corresponds to a physical place that may participate in workflow execution.

A site may register compute resources and data resources with NMDC. It may also execute workflows, and may request that workflows be executed.

A site must be able to service requests for any data resources it has registered.