Gladier (the “GLobus Architecture for Data-Intensive Experimental Research”) is a Python toolkit for developing data collection, analysis, and publication pipelines (“flows”) for experimental facilities. A flow might, for example:
Retrieve data from an instrument, verify its quality, extract metadata, and publish data+metadata to a catalog, or:
Collect data from a series of experiments, train a machine learning model, and deploy the model to an instrument.
In these and many other applications, Gladier makes it easy to specify what actions to perform, and where, and then to execute those actions reliably and securely.
Gladier builds on the powerful cloud-hosted Globus platform, including Globus automation services for reliable and scalable flow execution; Globus Auth for secure distributed operation; and services like Globus Transfer, Globus Compute, and Globus Search to implement data transfer, compute, cataloging, and other actions.
You can read more about Gladier, including example applications, in a technical report; here we focus on how to install and use it, and provide pointers to sample code.
- Running Flows
- Flow Generation
- Passing Payloads
- Auth in Gladier