Abstract. This report surveys two requirement-aware load scheduling policies: Requirement-Aware Load Selection (RALS) policy [1] and Minimum Slack Capacity First (MSCF) policy [2]. This report introduces a novel scheduling policy to overcome starvation in RALS and to deal with partially available prediction information available to the scheduler. As future work, the modified RALS may be combined with PAR (Prudent Algorithm with Replication) to form RAHAR (Requirement-Aware Hybrid Algorithm with Replication). For MSCF, no limitations nor drawbacks were found.
The survey was conducted as part of the Real-Time Systems class (EE5903) by Prof. Bharadwaj Veeravalli at the National University of Singapore in Spring 2017. The survey is available upon request.
References
[1] Hu, Menglan, and Bharadwaj Veeravalli. “Requirement-Aware Scheduling of Bag-of-Tasks Applications on Grids with Dynamic Resilience.” IEEE Transactions on Computers 62.10 (2013): 2108-2114.
[2] Hu, Menglan, and Bharadwaj Veeravalli. “Requirement-aware strategies for scheduling real-time divisible loads on clusters.” Journal of Parallel and Distributed Computing 73.8 (2013): 1083-1091.
Mike Hao Jiang 