ARCHIVED: What is co-scheduling, and does XSEDE support it?
Definition
Co-scheduling gives you simultaneous access to resources that are part of different logical systems and, potentially, in different physical locations.
Co-scheduling lets you run jobs that have resource requirements beyond those of any one resource. For example, co-scheduling resources lets you:
- Run jobs that require simultaneous access to a computational system and a visualization system
- Access a scientific instrument to perform visualization in parallel with a simulation
- Perform data analysis in parallel with data acquisition
Co-scheduling is distinct from on-demand computing, in which running or queued jobs may be preempted for time-critical applications (e.g., earthquake or tornado warning operations).
Implementation on XSEDE
Co-scheduling services are not yet widely in production on the Extreme Science and Engineering Discovery Environment (XSEDE), though they may be available experimentally on some resources.
Note: Co-scheduling is distinct from advance reservation. Advance reservation requests resources from a single logical system. Co-scheduling allocates resources from multiple logical systems. However, co-scheduling is often implemented atop advance reservations.
Getting help
If you need help or have questions about co-scheduling on XSEDE, contact the XSEDE Help Desk.
This is document avwz in the Knowledge Base.
Last modified on 2018-03-01 16:50:00.