We rolled out a number of smaller improvements, such as being able to label a workflow as active/inactive. This allows us to disable a configured workflow if a customer wishes to prevent that workflow from executing.
The Workflow API has been extended to support complex and nested parameter objects, allowing for greater configurability when designing flows. The workflow system now allows a flattened JSON representation of the configuration to be provided to steps in the workflow system, in place of the flat key-value approach previously supported.
We successfully migrated the Workflow system to our new cluster manager, allowing us to simplify our infrastructure and deployment models, and begin the decommissioning process for the old cluster manager, to be concluded in Q1 2018. This will allow for greater control over prioritisation and quotas, as well as the necessary data to integrate into the console views.
A number of improvements to performance and stability of the Workflow Service have been rolled out. Workflow is now capable of balancing execution across multiple clusters to improve overall throughput and pushing the theoretical performance limit of the platform towards 100,000 executions per-day.
Workflow Events now support localisation on the client side, allowing for consumers of the API to appropriately localise the events.
Additional telemetry and statistics have been added to the Workflow engine, allowing the platform to provide more detailed insight into the functioning of the Workflow service as a whole. This will help to drive further improvements.
Following a customer request, we have made improvements to the standard TVA ingest adapter and introduced new operational filtering in the Workflow Service for debugging and monitoring purposes. Now, it is possible to verify the status of an ingest execution by the file drop off location.
The Flow Language, which is used to define the execution steps in workflows, now has support for interacting with the Metadata Service to retrieve Contents and Assets, allowing other steps to make use of the metadata in decisions about how and what they produce.