Open Science Grid (OSG)
Open Science Grid (OSG) national, distributed computing resource available for free for running thousands of independent computations.
US-based research projects can run jobs via OSG Connect which is the online portal through which users can access their allocation. Main financial support for the OSG Team comes from NSF funded projects, with additional funding from several organizations.
Contents
Resource Description
Open Science Grid (OSG) is an NSF-funded distributed computing infrastructure for researchers affiliated with a U.S. institution. Free to use, this infrastructure provides a platform on which running thousands of independent computations is made convenient. Considered High-Throughput Computing (HTC), this resource benefits researchers who must handle a large quantity of independent tasks.
Eligibility
Any researcher affiliated with a US Academic Institution may request an allocation for free.
Use Case
Researchers with thousands of independent jobs that don’t require immense GPU power might consider the Open Science Pool (OSPool) for their work.
Getting Started
Researchers who wish to use this resource will register for an account on the OSPool, Jobs are managed via HTCondor, and researchers familiar with Slurm will find job scheduling similar. OSPool provides documentation on converting workflows from Slurm to HTCondor.
OSPool Quickstart
- Determine if your research project is a good fit for the OSPool. Review your project details and the table on OSG’s website.
- Researchers working with large amounts of data for each job, consider requesting space on the Open Science Data Federation (OSDF) by emailing OSG support. Space made available on the OSDF is accessible as a directory within your allocation on the OSPool.
- Create an account for OSG Connect and schedule a meeting with the OSG Research Facilitation team.
- OSG provides tutorials, available on Github, for researchers to familiarize themselves with the basics of workflow management on OSG.
- For further questions, OSG provides support via email, virtual office hours, and training sessions.
Support
Research Cyberinfrastructure is available to help researchers interested in using this resource through both the application and onboarding process.