Job communication characterization and its impact on meta-scheduling co-allocated jobs in a mini-grid

Jones, William M.; Pang, L.W.; Stanzione, Dan; Ligon, W.B.

doi:10.1109/ipdps.2004.1303317

Cited by 6 publications

(4 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Beosim [8] is a discrete event driven simulator designed to model a multi-cluster as a collection of (possibly heterogeneous) computational clusters connected via a dedicated interconnection network. Beosim can be driven via synthetic workload distributions that are characterized through the use of randomly generated arrival and service processes.…”

Section: Simulatormentioning

confidence: 99%

“…In this paper, we develop a parallel job model that takes both computation and communication into account as a means by which to explore co-allocating multi-cluster schedulers that exploit these unique architectural features [8]. We present an in-depth explanation of our communication model and its associated algorithms as well as a study of the impact of co-allocation in a multi-cluster as a function of job communication characteristics and scheduling routines.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Characterization of Bandwidth-Aware Meta-Schedulers for Co-Allocating Jobs Across Multiple Clusters

et al. 2005

View full text Add to dashboard Cite

In this paper, we present a bandwidth-centric job communication model that captures the interaction and impact of simultaneously co-allocating jobs across multiple clusters. We compare our dynamic model with previous research that utilizes a fixed execution time penalty for co-allocated jobs. We explore the interaction of simultaneously co-allocated jobs and the contention they often create in the network infrastructure of a dedicated computational multi-cluster.We also present several bandwidth-aware co-allocating meta-schedulers. These schedulers take inter-cluster network utilization into account as a means by which to mitigate degraded job run-time performance. We make use of a bandwidth-centric parallel job communication model that captures the time-varying utilization of shared inter-cluster network resources. By doing so, we are able to evaluate the performance of multi-cluster scheduling algorithms that focus not only on node resource allocation, but also on shared inter-cluster network bandwidth.

show abstract

Section: Simulatormentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Characterization of Bandwidth-Aware Meta-Schedulers for Co-Allocating Jobs Across Multiple Clusters

et al. 2005

View full text Add to dashboard Cite

show abstract

“…DAS is a homogeneous minigrid environment containing 200 nodes in total. The Parallel Architecture Research Laboratory (PARL) at Clemson University has also set up a computational minigrid consisting of totally 792 homogeneous Pentium III processors, which belong to five Beowulf Clusters interconnected through dedicated Ether links [35], [36]. 3.…”

Section: Minigrid Environments Consideredmentioning

confidence: 99%

Agent-based load balancing on homogeneous minigrids: macroscopic modeling and characterization

Liu

Jin

Wang³

2005

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

Abstract-In this paper, we present a macroscopic characterization of agent-based load balancing in homogeneous minigrid environments. The agent-based load balancing is regarded as agent distribution from a macroscopic point of view. We study two quantities on minigrids: the number and size of teams where agents (tasks) queue. In macroscopic modeling, the load balancing mechanism is characterized using differential equations. We show that the load balancing we concern always converges to a steady state. Furthermore, we show that load balancing with different initial distributions converges to the same steady state gradually. Also, we prove that the steady state becomes an even distribution if and only if agents have complete knowledge about agent teams on minigrids. Utility gains and efficiency are introduced to measure the quality of load balancing. Through numerical simulations, we discuss the utility gains and efficiency of load balancing in different cases and give a series of analysis. In order to maximize the utility gain and the efficiency, we theoretically discuss the optimization of agents' strategies. Finally, in order to validate our proposed agentbased load balancing mechanism, we develop a computing platform, called Simulation System for Grid Task Distribution (SSGTD). Through experimentation, we note that our experimental results in general confirm our theoretical proofs and numerical simulation results on the proposed equation system. In addition, we find a very interesting phenomenon, that is, our agent-based load balancing mechanism is topology-independent.Index Terms-Homogeneous minigrids, load balancing, task distribution, agents, macroscopic modeling, steady states, convergence, grid simulation. Since a grid connects numerous geographically distributed computers, and tasks are submitted to grid nodes in a distributed fashion, an important issue is how to evenly distribute submitted tasks to nodes. This is a load balancing problem, one of the scheduling problems on the grid. By solving this problem, we can optimally utilize computational resources of the grid. In this paper, we will propose an agent-based load balancing mechanism. Scheduling on GridsThe scheduling problem on grids has been widely studied [6] have studied the task allocation problem in a grid environment, where the submitted task is arbitrarily divisible. In other words, a task can be divided into arbitrary chunks. Their scheduling mechanism assumed a master/worker architecture, i.e., a master, acting as a scheduler, is responsible for dividing the submitted tasks and allocating the obtained task chunks to different workers. In their work, they paid special attention to task transfer time and network latency. Their algorithm, called uniform multiround (UMR), allocates chunks using multiple rounds so as to overlap communication (i.e., transfer time and network latency) and computation and, consequently, decrease the total time for handling the task, i.e., makespan.In Casanova et al.'s work, the main problem lies in the master/wor...

show abstract

“…Moreover, the slowdown ratio may actually change dynamically based on the jobs' communication patterns, workload on the network links, and the maximum bandwidth of the network links. Instead of using a static communication model, other researchers [9], tried to use a more dynamic view of job communication that is bandwidth-centric to analyze their performance characteristics. This work has been limited to jobs with all-to-all global communication patterns.…”

Section: Introductionmentioning

confidence: 99%

A Study on Job Co-Allocation in Multiple HPC Clusters

Qin

Bauer

2006

20th International Symposium on High-Performance Computing in an Advanced Collaborative Environment (HPCS'06)

View full text Add to dashboard Cite

To more effectively use HPC clusters for even larger computations, improve turn-around times and better utilize compute resource, users are looking to interconnect multiple HPC clusters, creating a grid. To effectively use such grids, it may be desirable to split and co-allocate jobs requiring many processes across multiple clusters. While splitting a very large job across multiple clusters is an attractive possibility, the benefit, in terms of improving turn-around time, ultimately depends on the communication patterns between processes, workload on the communication links, and the maximum bandwidth of the links. The objective of this work is to understand the impact of communications on multi-processor jobs in order to develop scheduling strategies and job allocation algorithms for multi-cluster grids which can accommodate communication factors. In this paper we report on initial investigations of some co-allocation strategies. This evaluation is based on a simulator that has been implemented and validated experimentally across two HPC clusters.

show abstract

Job communication characterization and its impact on meta-scheduling co-allocated jobs in a mini-grid

Cited by 6 publications

References 9 publications

Characterization of Bandwidth-Aware Meta-Schedulers for Co-Allocating Jobs Across Multiple Clusters

Characterization of Bandwidth-Aware Meta-Schedulers for Co-Allocating Jobs Across Multiple Clusters

Agent-based load balancing on homogeneous minigrids: macroscopic modeling and characterization

A Study on Job Co-Allocation in Multiple HPC Clusters

Contact Info

Product

Resources

About