Power and Thermal-Aware Workload Allocation in Heterogeneous Data Centers

Al-Qawasmeh, Abdulla M.; Pasricha, Sudeep; Maciejewski, Anthony A.; Siegel, Howard Jay

doi:10.1109/tc.2013.116

Cited by 57 publications

(26 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Examples of prior work in scheduling for large-scale HPC systems have appeared in [15][16][17][18]. In [15], the authors look at the problem of energy-constrained dynamic allocations of tasks in heterogeneous cluster computing environments, in the presence of individual task deadlines.…”

Section: Scheduling Heuristicsmentioning

confidence: 99%

“…In [15], the authors look at the problem of energy-constrained dynamic allocations of tasks in heterogeneous cluster computing environments, in the presence of individual task deadlines. The work in [16] proposes power and thermal-aware scheduling to optimize individual tasks reaching their deadlines. A "utility" metric is defined in [17] that is used in combination with several energy-aware scheduling heuristics to provide a method of resource allocation that can maximize task performance while operating under a system energy constraint.…”

Section: Scheduling Heuristicsmentioning

confidence: 99%

“…If the calculated slack prediction is greater than zero, then N A is set to true, the available node n j is recorded to the variable that stores the node with available slack N W AS, and the algorithm breaks from the inner loop, otherwise it checks the next processor node (lines 9-12). Once the inner for loop is complete, if there was a node available, then task u i is scheduled onto the node recorded in N W AS, otherwise the algorithm moves on the next unmapped task (lines [15][16]. After the outer for loop has iterated through all the unmapped tasks, the algorithm is complete.…”

Section: Co-location Naïve Scheduling Heuristicmentioning

confidence: 99%

See 2 more Smart Citations

HPC node performance and energy modeling with the co-location of applications

et al. 2016

Self Cite

View full text Add to dashboard Cite

Multicore processors have become an integral part of modern large-scale and high-performance parallel and distributed computing systems. Unfortunately, applications co-located on multicore processors can suffer from decreased performance and increased dynamic energy use as a result of interference in shared resources, such as memory. As this interference is difficult to characterize, assumptions about application execution time and energy usage can be misleading in the presence of colocation. Consequently, it is important to accurately characterize the performance and energy usage of applications that execute in a co-located manner on these architec-tures. This work investigates some of the disadvantages of co-location, and presents a methodology for building models capable of utilizing varying amounts of information about a target application and its co-located applications to make predictions about the target application's execution time and the system's energy use under arbitrary co-locations of a wide range of application types. The proposed methodology is validated on three different server class Intel Xeon multicore processors using eleven applications from two scientific benchmark suites. The model's utility for scheduling is also demonstrated in a simulated large-scale high-performance computing environment through the creation of a co-location aware scheduling heuristic. This heuristic demonstrates that scheduling using information generated with the proposed modeling methodology is capable of making significant improvements over a scheduling heuristic that is oblivious to co-location interference.

show abstract

Section: Scheduling Heuristicsmentioning

confidence: 99%

Section: Scheduling Heuristicsmentioning

confidence: 99%

Section: Co-location Naïve Scheduling Heuristicmentioning

confidence: 99%

See 1 more Smart Citation

HPC node performance and energy modeling with the co-location of applications

et al. 2016

Self Cite

View full text Add to dashboard Cite

show abstract

“…Khemka et al [12] propose a method that iteratively increases the TMA of an existing matrix while keeping the same MPH and TDH. A method that generates matrices with varying affinities (similar to the TMA) and which resembles the noise-based method is also proposed in [3]. However, no formal method has been proposed for generating matrices with a given TMA.…”

Section: Related Workmentioning

confidence: 99%

Controlling and Assessing Correlations of Cost Matrices in Heterogeneous Scheduling

Canon

Héam

Philippe

2016

Euro-Par 2016: Parallel Processing

View full text Add to dashboard Cite

Abstract. This paper considers the problem of allocating independent tasks to unrelated machines such as to minimize the maximum completion time. Testing heuristics for this problem requires the generation of cost matrices that specify the execution time of each task on each machine. Numerous studies showed that the task and machine heterogeneities belong to the properties impacting heuristics performance the most. This study focuses on orthogonal properties, the average correlations between each pair of rows and each pair of columns, which is a proximity measure with uniform instances 1 . Cost matrices generated with a novel generation method show the effect of these correlations on the performance of several heuristics from the literature. In particular, EFT performance depends on whether the tasks are more correlated than the machines and HLPT performs the best when both correlations are close to one.

show abstract

“…The research work proposed in [15] confronts issues pertaining to maximizing the performance of a data center taking into consideration the deadline of competing tasks.The study also addresses issues concerning reduction in power consumption and thermal constraints. Thermal aware resource allocation techniques are developed.…”

mentioning

confidence: 99%

Energy-Aware Scheduling Framework for resource allocation in a virtualized cloud data centre

Gupta¹,

Katiyar²

2017

IJET

View full text Add to dashboard Cite

Abstract-Cloud paradigm is an embryonic computing model that in its vicinity stresses on proficient utilization of computing resources. Data centers that host and service cloud applications ingest enormous amount of energy, leading to massive emission of carbon footprints to the atmosphere and high operational expenditures. Consequently, there is a need to establish synergy between data centre resources for optimum resource utilization and strategies needs to be devised that can considerably reduce energy consumption in cloud data center. This paper elucidates an architectural framework for computation of energy spent in scheduling resources on hosts. The framework has been implemented for bin-packing techniques and explicates minutiae about broker components involved in scheduling process. Keyword-Cloud Computing, Energy Consumption, VM Migration, Resource Scheduling.I. INTRODUCTION Contemporary [1,2] resource-intensive enterprises has engendered demands for high performance computing infrastructures. Proliferation of IT services to be used by diverse range of cloud users has led to construction of large-scale energy hungry data centers that can facilitate computing services. Despite of the improvisations introduced in energy consumption models, service providers are confronted with challenges of reduction in energy consumption and CO 2 emission.The rationale [3] in the wake of explosion of energy emission is increase in number of computer usage due to increase in number of IT practitioners. As an upshot, size of data centres has increased. Moreover, exploiting energy-aware resource provisioning to its fullest extent can subsequently provide a solution to the forefront issues.[4] Service virtualization and consolidation are acting as inherent practices that can escort energyefficient datacenter architectures. It has effectually led to efficient resource utilization. VM provisioning can be sighted as a multidimensional bin packing problem comprising of capricious bin configurations and cost parameters. Virtualization technique embraces server consolidation process and a VM live-migration technique that has validated to be efficient in drastic reduction in energy consumption in high-performance cloud datacenters. However, I/O virtualization has excavated grounds for performance degradation posed by overheads encountered in vm migrations and needs to be addressed urgently.The organization of paper is laid out as literature review in beginning followed by research focus. The IV section elucidates architectural framework trailed by working prototype. The last section illustrates conclusion and future work.II. LITERATURE REVIEW The research work presented in [5] explores method to manage data intensive distributed programming paradigms (like MapReduce and Dryad) that assists practitioners to effortlessly parallelize the processing of huge data sets. Deployment of such data intensive computing infrastructures is of significant concern due to rise in cost. The work carried out in the study dynamically adjusts the size ...

show abstract

Power and Thermal-Aware Workload Allocation in Heterogeneous Data Centers

Cited by 57 publications

References 42 publications

HPC node performance and energy modeling with the co-location of applications

HPC node performance and energy modeling with the co-location of applications

Controlling and Assessing Correlations of Cost Matrices in Heterogeneous Scheduling

Energy-Aware Scheduling Framework for resource allocation in a virtualized cloud data centre

Contact Info

Product

Resources

About