OpenScienceGrid (OSG)

Overview

The Open Science Grid (OSG) is "a national, distributed computing partnership for data-intensive research" [0].

Creating an account is free, and access to OSG's resources provides researchers with an expansive (and free) computing pool.

See the accounting portal for an overview of OSG's total compute time (700 million compute hours from 2016-01-06 to 2019-06-01) [1].

To get started, first make sure you have an activated Globus account (follow the signup instructions here [2]). Note that this process involves a phone interview. 

OSG provides a "Quickstart" guide here [3]. Before users can remotely access the OSG login nodes with ssh, an ssh-key needs to be added to the user's Globus profile. This may take overnight to synchronize.

Globus provides documentation on how to generate these keys and how to add them to the Globus ID profile [4]. After the keys are properly set up and synchronized, the login command is

ssh your_username@login.osgconnect.net

Additional login nodes may be accessed (the above domain defaults to one of these):

  • login01.osgconnect.net

  • login02.osgconnect.net

  • login03.osgconnect.net

There are additional login nodes but these three are sufficient. At the time of this writing (2019-06-13) login03.osgconnect.net is the default login node.

 

Tutorials

  • An interactive tutorial from the University of Utah.

    • This resource provides a guide to getting started with and using OSG (and its queuing software HTCondor).

    • OSG's login nodes provide tutorial programs that users may run within the shell. The guide runs through several of these, getting more complex with each example. The dag files and submit scripts provided serve as decent templates for new users.

  • User Guide for OSG.

    • This resource provides good computational guidelines as well as some sample scripts

OSG and ASU

ASU hosts five OSG compute nodes within the Sol Supercomputer!

Additional Help