Grid-enabling applications in a heterogeneous environment with globus and condor

The authors Grid-enabled several test applications to conduct experiments that used the Condor Job Manager to schedule code execution across both Globus and Condor resources. Using Condor to manage Globus Grid services in addition to local Condor pools (collections of workstations) increases scheduling flexibility for assigning jobs to the optimal processing locations. In universities, workstations in computer labs can be harnessed when they are closed to students. In other enterprises, employee workstations can become a powerful HPC after work hours. We are also currently implementing a Globus Global Resource Access Manager (GRAM) for a Linux cluster that will also be managed by Condor to bring a third class of system into our heterogeneous environment.