Resource Allocation Policies for Loosely Coupled Applications in Heterogeneous Computing Systems

Hwang, Eunji; Kim, Suntae; Yoo, Tae-kyung; Kim, Jik-Soo; Hwang, Soonwook; Choi, Young-ri

doi:10.1109/tpds.2015.2461154

Cited by 22 publications

(17 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CPU cores or CPU cycles) equally to applications submitted by the users. For a heterogeneous computing system composed of different computing platforms, we can simply allocate the equal number of cores from each platform to each application or user 1 to achieve the fairness as discussed in [20]. However, depending on the characteristics of each application, the application has different sets of preferred platforms and co-runners.…”

Section: Fairness In Heterogeneous Computing Systemsmentioning

confidence: 99%

“…For all the metrics below, a higher metric value means that it is more beneficial to run an application K on a platform P , rather than the other platforms in the system. Throughput A throughput of a many-task application on a platform with OneCore (i.e., the number of tasks processed using one core per hour) can be used for platform affinity as in [20], [22]. It is a simple raw metric to show the performance of the application on that particular platform.…”

Section: Platform Affinity Metricmentioning

confidence: 99%

“…For each submitted application, the system creates a separate queue to maintain its tasks. For an allocated core to the application, a pilot job is deployed to continuously fetch and execute a task of the application from the queue until there is no task left in the queue [6], [16], [20], [23].…”

Section: Two-level Scheduling Algorithmmentioning

confidence: 99%

See 2 more Smart Citations

Achieving Fairness-Aware Two-Level Scheduling for Heterogeneous Distributed Systems

Hwang

Kim

Choi

2021

IEEE Trans. Serv. Comput.

Self Cite

View full text Add to dashboard Cite

In a heterogeneous distributed system composed of various types of computing platforms such as supercomputers, grids, and clouds, a two-level scheduling approach can be used to effectively distribute resources of the platforms to users in the first-level, and map tasks of the users in nodes for each platform in the second-level for executing many-task applications. When scheduling heterogeneous resources, service providers of the system should consider the fairness among multiple users as well as the system efficiency. However, the fairness cannot be achieved by simply distributing an equal amount of resources from each platform to every user. In this paper, we investigate how to address the fairness issue among multiple users in a heterogeneous distributed system. We present three first-level resource allocation policies of a provider affinity first policy, an application affinity first policy, and a platform affinity based round-robin policy, and two second-level task mapping policies of a most affected first policy and a co-runner affinity based round-robin policy. Using trace-based simulations, we evaluate the performance of various combinations of the first and second level scheduling policies. Our extensive simulation results demonstrate that the first-level policy plays a crucial role to achieve relatively good fairness.

show abstract

Section: Fairness In Heterogeneous Computing Systemsmentioning

confidence: 99%

Section: Platform Affinity Metricmentioning

confidence: 99%

See 1 more Smart Citation

Achieving Fairness-Aware Two-Level Scheduling for Heterogeneous Distributed Systems

Hwang

Kim

Choi

2021

IEEE Trans. Serv. Comput.

Self Cite

View full text Add to dashboard Cite

show abstract

“…There are several inspiring load balancing approaches which have presented interesting results. Some of them are proposed for distributed systems [8], grid [9], [10], P2P systems [11], [12], and also for cloud computing systems [13], [14], and heterogeneous computing systems [15], [16], [17]. The main idea behind the considered challenge is the effective method for task assignment.…”

Section: Related Workmentioning

confidence: 99%

A Mobile Agent Team Works based on Load-Balancing Middleware for Distributed Computing Systems

Benchara¹,

Youssfi²

2019

IJACSA

View full text Add to dashboard Cite

The aim of this paper is to present a load balancing middleware for parallel and distributed systems. The great challenge is to balance the tasks between heterogeneous distributed nodes for parallel and distributed computing models based distributed systems, by the way to ensure HPC (High performance computing) of these models. Accordingly, the proposed middleware is based on mobile agent team work which implements an efficient method with two strategies: (i) Load balancing Strategy that determines the node tasks assignment based on node performance, and (ii) Rebalancing Strategy that detects the unbalanced nodes and enables tasks migration. The paper focuses on the proposed middleware and its cooperative mobile agent team work model strategies to dynamically balance the nodes, and scale up distributed computing systems. Indeed, some experimental results that highlight the performance and efficiency of the proposed middleware are presented.

show abstract

“…Applications with workloads comprised of multiple tasks impose sophisticated execution and advanced resource management requirements [1]. Highperformance computing (HPC) systems have been designed to support applications comprised of mostly monolithic, single-job workloads.…”

Section: Introductionmentioning

confidence: 99%

Using Pilot Systems to Execute Many Task Workloads on Supercomputers

Merzky

Turilli

Maldonado

et al. 2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

High performance computing systems have historically been designed to support applications comprised of mostly monolithic, singlejob workloads. Pilot systems decouple workload specification, resource selection, and task execution via job placeholders and late-binding. Pilot systems help to satisfy the resource requirements of workloads comprised of multiple tasks. RADICAL-Pilot (RP) is a modular and extensible Python-based pilot system. In this paper we describe RP's design, architecture and implementation, and characterize its performance. RP is capable of spawning more than 100 tasks/second and supports the steady-state execution of up to 16K concurrent tasks. RP can be used stand-alone, as well as integrated with other application-level tools as a runtime system.

show abstract

Resource Allocation Policies for Loosely Coupled Applications in Heterogeneous Computing Systems

Cited by 22 publications

References 26 publications

Achieving Fairness-Aware Two-Level Scheduling for Heterogeneous Distributed Systems

Achieving Fairness-Aware Two-Level Scheduling for Heterogeneous Distributed Systems

A Mobile Agent Team Works based on Load-Balancing Middleware for Distributed Computing Systems

Using Pilot Systems to Execute Many Task Workloads on Supercomputers

Contact Info

Product

Resources

About