Performance modeling of parallel applications for grid scheduling

Sanjay, H. A.; Vadhiyar, Sathish

doi:10.1016/j.jpdc.2008.02.006

Cited by 37 publications

(31 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results also show that the predicted execution times shown in Figure 7(b) are highly accurate and are within 10-30% of the actual execution times shown in Figure 7(a). This confirms the conclusions in our previous work [45] regarding the accuracy of the performance models. We also find that the relative differences between the predicted execution times of the schedules for the different algorithms, as shown in Figure 7(b), match the relative differences between the actual execution times shown in Figure 7(a).…”

Section: Real Experimentssupporting

confidence: 92%

“…In our previous work [45], we developed performance modeling strategies for predicting the execution times of tightly coupled parallel applications on non-dedicated homogeneous resources. We calculated the time taken for the execution of a parallel application as: The formula shown in Equation (1) splits the execution time of a parallel application into two parts, f comp and f comm , for representing computation and communication aspects, respectively, of the parallel application.…”

Section: Application Performance Modelsmentioning

confidence: 99%

“…We also devised a strategy for scaling the coefficients of computational complexities for predicting the execution time of a parallel application on a cluster using the performance models developed for the application in another cluster. The details of the performance models, the modeling procedure and the results can be found in our previous work [45].…”

Section: Application Performance Modelsmentioning

confidence: 99%

See 2 more Smart Citations

A strategy for scheduling tightly coupled parallel applications on clusters

Sanjay

Vadhiyar

2009

Concurrency and Computation

View full text Add to dashboard Cite

SUMMARYAlthough various strategies have been developed for scheduling parallel applications with independent tasks, very little work exists for scheduling tightly coupled parallel applications on cluster environments. In this paper, we compare four different strategies based on performance models of tightly coupled parallel applications for scheduling the applications on clusters. In addition to algorithms based on existing popular optimization techniques, we also propose a new algorithm called Box Elimination that searches the space of performance model parameters to determine the best schedule of machines. By means of real and simulation experiments, we evaluated the algorithms on single cluster and multi-cluster setups. We show that our Box Elimination algorithm generates up to 80% more efficient schedules than other algorithms. We also show that the execution times of the schedules produced by our algorithm are more robust against the performance modeling errors.

show abstract

Section: Real Experimentssupporting

confidence: 92%

Section: Application Performance Modelsmentioning

confidence: 99%

Section: Application Performance Modelsmentioning

confidence: 99%

See 1 more Smart Citation

A strategy for scheduling tightly coupled parallel applications on clusters

Sanjay

Vadhiyar

2009

Concurrency and Computation

View full text Add to dashboard Cite

show abstract

“…• Code analysis (Nudd et al, 2000;Reistad and Gifford, 1994) • Analytic benchmarking/code profiling (Yang et al, 1993) • Historical/Statistical prediction (Sanjay and Vadhiyar, 2008;Iverson et al, 1996) • Empirical analysis (Berman et al, 2005) Using these solutions, or by executing part of the code, other information related to user jobs can be obtained. In addition, there exists many studies which proposed models to estimate specific parameters (e.g., Cycle Per Instruction (CPI) (Chen and John, 2011;Intel Corporation, 2008), Memory Access Per Instruction (MPI) (Zhang and Chang, 2014) and estimated bandwidth (Zhu et al, 2012).…”

Section: Profiling and Prediction Phasementioning

confidence: 99%

Towards Solving the Problem of Virtual Machine Placement in Cloud Computing: A Job Classification Approach

Al-Dulaimy¹,

Zekri²,

Itani³

et al. 2016

Journal of Computer Science

View full text Add to dashboard Cite

Cloud Computing is a paradigm that delivers services by providing an access to wide range of shared resources which are hosted in cloud data centers. One of the recent challenges in this paradigm is to enhance the energy efficiency in these data centers. In this study, a model that identifies common patterns for the jobs submitted to the cloud is proposed. This model is able to predict the type of the job submitted and accordingly, the set of users' jobs is classified into four subsets. Each subset contains jobs that have similar requirements. In addition to the jobs' common pattern and requirements, the users' history is considered in the jobs' type prediction model. The goal of job classification is to find a way to propose useful strategy that helps to improve power efficiency. Based on the process of jobs' classification, the best fit virtual machine is allocated to each job. Then, the virtual machines are placed on the physical machines according to a novel strategy, called Mixed Type Placement strategy. The core idea of the proposed strategy is to place virtual machines of the jobs of different types in the same physical machine whenever possible. The placement process is based on Multi Choice Knapsack Problem which is a generalization of the classical Knapsack Problem. This is because different types of jobs do not intensively use the same compute or storage resources in the physical machine. This strategy minimizes the number of active physical machines which, in turn, leads to major reduction in the total energy consumption in the data center. The total execution time and the cost of executing the jobs submitted are considered in the placement process. To evaluate the performance of the proposed strategy, the CloudSim simulator is used with a real workload trace to simulate the cloud computing environment. The results show that the proposed strategy outperform both Genetic Algorithm and Round Robin from energy efficiency perspective.

show abstract

“…Emulation demonstrates the conditions for which, splitting a TCP connection is most useful and reveals a significant source of overhead. In [19], Sanjay et al developed a comprehensive set of performance modeling strategies for predicting execution times of parallel applications on both dedicated and non-dedicated environments. These strategies adapt to changing network and CPU loads on the grid resources.…”

Section: Related Workmentioning

confidence: 99%