Grids are made of heterogeneous computing resources geographically dispersed where providing QoS is a challenging task. One way of enhancing the QoS perceived by users is by performing scheduling of jobs in advance, since reservations of resources are not always possible. In this way, it becomes more likely that the appropriate resources are available to run the job when needed. However, as resources in Grid environments may be shared among Grid users and local users, monitoring and prediction techniques are needed to estimate future durations of jobs.

The SA-Layer is a meta-scheduling framework designed by the GridCloud team at the I3A (UCLM) to provide the functionality needed to perform meta-scheduling of jobs in advance, which means that scheduling decisions are performed some time before jobs are actually executed. To this end:

  • It is implemented as a layer on top of GridWay Meta-scheduler and Globus Toolkit (GT4).
  • It uses functionality provided by GridWay in terms of resource discovery and monitoring, job submission and execution monitoring, etc.
  • It divides resource usage into time slots and schedules the future usage of them.
  • It keeps track of previous decisions in order to not overlap executions.
  • It makes estimations for job durations by estimating the execution time of a job and the time needed to complete data transfers separately.
  • It makes predictions about the future status of Grid resources, network inclusive.
  • It estimates how predictable the resource behavior is.

The main aim of SA-Layer is to ensure resource availability for future job executions and to provide QoS to end-users in terms of finished jobs fulfilling deadlines.

Home Page Image