Evaluation of an Adaptive Scheduling Strategy for Master-Worker Applications on Clusters of Workstations

Efficient resource management applied to master–worker applications

“…1, which shows the effect on efficiency and execution time of a largest processing time first (LPTF) policy. These results were obtained in a simulation study comparing different scheduling strategies for master-worker applications, which can be found in [10]. In general, our results showed that for any given workload distribution, a similar scenario to the one depicted in Fig.…”

“…The first of these controls the assignment of tasks to workers according to an algorithm that keeps a running average of each task (a detailed description could be found in [10]). Tasks are sorted in decreasing order of this average execution time and, at each iteration, were assigned to workers according to the sorted list.…”

Bandwidth-aware resource allocation for heterogeneous computing systems to maximize throughput

“…The Condor project [14] develops a software infrastructure so that heterogeneous resources with distributed ownerships can be utilized to provide large amounts of processing capacity over long periods of time. Master-slave paradigm is widely used in the Condor systems and has been exploited by various research efforts ( [6,13]) to maximize the throughput. The multi-level master-slave paradigm is studied in [1], where a bandwidth-centric approach was proposed to maximize system throughput.…”

Evolutionary Optimization Techniques on Computational Grids

“…However, our optimization framework is based on an iterative way of working. For this purpose, we have developed our optimization framework based on an extended version of MW [7], which allows MW to iterate a predetermined number of times.…”