Design and implementation of an adaptive job allocation strategy for heterogeneous multi‐cluster computing systems

Cluster computing is an attractive approach to provide high‐performance computing for solving large‐scale applications. Owing to the advances in processor and networking technology, expanding clusters have resulted in the system heterogeneity; thus, it is crucial to dispatch jobs to heterogeneous computing resources for better resource utilization. In this paper, we propose a new job allocation system for heterogeneous multi‐cluster environments named the Adaptive Job Allocation Strategy (AJAS), in which a self‐scheduling scheme is applied in the scheduler to dispatch jobs to the most appropriate computing resources. Our strategy focuses on increasing resource utility by dispatching jobs to computing nodes with similar performance capacities. By doing so, execution times among all nodes can be equalized. The experimental results show that AJAS can improve the system performance. Copyright © 2011 John Wiley & Sons, Ltd.     

Security-Aware Resource Allocation for Real-Time Parallel Jobs on Homogeneous and Heterogeneous Clusters

Security is increasingly becoming an important issue in the design of real-time parallel applications, which are widely used in the industry and academic organizations. However, existing resource allocation schemes for real-time parallel jobs on clusters generally do not factor in security requirements when making allocation and scheduling decisions. In this paper, we develop two resource allocation schemes, called task allocation for parallel applications with deadline and security constraints (TAPADS) and security-aware and heterogeneity-aware resource allocation for parallel jobs (SHARP), by taking into account applications' timing and security requirements in addition to precedence constraints. We consider two types of computing platforms: homogeneous clusters and heterogeneous clusters. To facilitate the presentation of the new schemes, we build mathematical models to describe a system framework, security overhead, and parallel applications with deadline and security constraints. The proposed schemes are applied to heuristically find resource allocations that maximize the quality of security and the probability of meeting deadlines for parallel applications running on clusters. Extensive experiments using real-world applications and traces, as well as synthetic benchmarks, demonstrate the effectiveness and practicality of the proposed schemes.     

A new step toward load balancing based on competency rank and transitional phases in Grid networks

Grid computing is a network of software-hardware capabilities. It serves as a comprehensive and complete system for organizations by which the maximum utilization from resources is achieved. Resource distribution in a heterogeneous and unstable environment and also effective load distribution among these resources are the important and difficult problems in Grid networks. Using dynamic and static algorithms or searching tree and Branch and Bound algorithm are considered to be among the available methods to reach the load balancing in Grid networks. This paper presents a new method for dynamic load balancing. In this method, we use the subtraction of forward and backward ants as a competency rank to take the priority of the sites, and we use a control word to search the suitable resource as well. Our main purpose is to devote jobs to the existing resources based on their processing power. Simulation results show that the proposed method can reduce the total completion time and also total tardiness to get the load balancing. The cost of using resources as an effective factor in load balancing is also observed.     

On the performance-driven load distribution for heterogeneous computational grids

Load balancing has been a key concern for traditional multiprocessor systems. The emergence of computational grids extends this challenge to deal with more serious problems, such as scalability, heterogeneity of computing resources and considerable transfer delay. In this paper, we present a dynamic and decentralized load balancing algorithm for computationally intensive jobs on a heterogeneous distributed computing platform. The time spent by a job in the system is considered as the main issue that needs to be minimized. Our main contributions are: (1) Our algorithm uses site desirability for processing power and transfer delay to guide load assignment and redistribution, (2) Our transfer and location policies are a combination of two specific strategies that are performance driven to minimize execution cost. These two policies are the Instantaneous Distribution Policy (IDP) and the Load Adjustment Policy (LAP), (3) The communication overhead involved in information collection is reduced using mutual information feedback. The simulation results show that our proposed algorithm outperforms conventional approaches over a wide range of system parameters.     

Dynamic grid load sharing with adaptive dissemination protocols

Scheduling in large scale dynamic grids comprising eclectic collections of resources is increasingly difficult. Autonomous resource neighborhoods may wish to determine the level of grid offered load that they can or will accept; different sites may wish to attract different amounts of load, to satisfy some desired property within a grid economy. This changes the traditional notion of load sharing, which generally assumes that the desired equilibrium should be an equal distribution of load across all participating machines, because they are under the jurisdiction of a single site, and therefore more likely to implement one common policy. In large-scale grids, nodes and neighborhoods should instead get a portion of the load that best matches their local policies for supporting and admitting grid jobs. This article describes information dissemination protocols that can distribute load in this way, without using load rebalancing through job migration, which is more difficult and costly in large-scale heterogeneous grids. Essentially, nodes adjust their advertising rates and aggressiveness to influence where jobs get scheduled. We report experimental results with example resource configurations in which each resource neighborhood determines its ideal grid load and disseminates accordingly. In turn, each neighborhood attracts the requisite amount of resource requests from the grid. Moreover, performance does not degrade: overall query satisfaction rates are within 9% of both adaptive dissemination protocols that use static adaptation policies, and static dissemination protocols that may be custom-tailored to specific resource and load distributions.     

Double auction-inspired meta-scheduling of parallel applications on global grids

Meta-schedulers map jobs to computational resources that are part of a Grid, such as clusters, that in turn have their own local job schedulers. Existing Grid meta-schedulers either target system-centric metrics, such as utilisation and throughput, or prioritise jobs based on utility metrics provided by the users. The system-centric approach gives less importance to users’ individual utility, while the user-centric approach may have adverse effects such as poor system performance and unfair treatment of users. Therefore, this paper proposes a novel meta-scheduler, based on the well-known double auction mechanism that aims to satisfy users’ service requirements as well as ensuring balanced utilisation of resources across a Grid. We have designed valuation metrics that commodify both the complex resource requirements of users and the capabilities of available computational resources. Through simulation using real traces, we compare our scheduling mechanism with other common mechanisms widely used by both existing market-based and traditional meta-schedulers. The results show that our meta-scheduling mechanism not only satisfies up to 15% more user requirements than others, but also improves system utilisation through load balancing.     

Enhanced GridSim architecture with load balancing

Grid is a network of computational resources that may potentially span many continents. Maximization of the resource utilization hinges on the implementation of an efficient load balancing scheme, which provides (i) minimization of idle time, (ii) minimization of overloading, and (iii) minimization of control overhead. In this paper, we propose a dynamic and distributed load balancing scheme for grid networks. The distributed nature of the proposed scheme not only reduces the communication overhead of grid resources but also cuts down the idle time of the resources during the process of load balancing. We apply the proposed load balancing approach on Enhanced GridSim in order to gauge the effectiveness in terms of communication overhead and response time reduction. We show that significant savings are delivered by the proposed technique compared to other approaches such as centralized load balancing and no load balancing.     

Decentralized resource management for a distributed continuous media server

Distributed continuous media server (DCMS) architectures are proposed to minimize the communication-storage cost for those continuous media applications that serve a large number of geographically distributed clients. Typically, a DCMS is designed as a pure hierarchy (tree) of centralized continuous media servers. In an earlier work, we proposed a redundant hierarchical topology for DCMS networks, termed RedHi, which can potentially result in higher utilization and better reliability over pure hierarchy. We focus on the design of a resource management system for RedHi that can exploit the resources of its DCMS network to achieve these performance objectives. Our proposed resource management system is based on a fully decentralized approach to achieve optimal scalability and robustness. In general, the major drawback of a fully decentralized design is the increase in latency time and communication overhead to locate the requested object. However, as compared to the typically long duration and high resource/bandwidth requirements of continuous media objects, the extra latency and overhead of a decentralized resource management approach become negligible. Moreover, our resource management system collapses three management tasks: (1) object location, (2) path selection, and (3) resource reservation, into one fully decentralized object delivery mechanism, reducing the latency even further. In sum, decentralization of the resource management satisfies our scalability and robustness objectives, whereas collapsing the management tasks helps alleviate the latency and overhead constraints.     

Utopia: A load sharing facility for large,heterogeneous distributed computer systems

Load sharing in large, heterogeneous distributed systems allows users to access vast amounts of computing resources scattered around the system and may provide substantial performance improvements to applications. We discuss the design and implementation issues in Utopia, a load sharing facility specifically built for large and heterogeneous systems. The system has no restriction on the types of tasks that can be remotely executed, involves few application changes and no operating system change, supports a high degree of transparency for remote task execution, and incurs low overhead. The algorithms for managing resource load information and task placement take advantage of the clustering nature of large-scale distributed systems; centralized algorithms are used within host clusters, and directed graph algorithms are used among the clusters to make Utopia scalable to thousands of hosts. Task placements in Utopia exploit the heterogeneous hosts and consider varying resource demands of the tasks. A range of mechanisms for remote execution is available in Utopia that provides varying degrees of transparency and efficiency. A number of applications have been developed for Utopia, ranging from a load sharing command interpreter, to parallel and distributed applications, to a distributed batch facility. For example, an enhanced Unix command interpreter allows arbitrary commands and user jobs to be executed remotely, and a parallel make facility achieves speed-ups of 15 or more by processing a collection of tasks in parallel on a number of hosts.     

气象环境下的云计算调度算法

针对气象计算的特点,提出气象计算的云模型,在这个模型之上,提出气象云计算（Weather Cloud）的启发式调度算法。调度算法对气象作业按照时间紧迫型、CPU紧迫型、内存紧迫型和硬盘空间紧迫型进行分类,计算资源综合紧迫指数,相应地赋予不同调度优先权限。与CMMS（Cloud Min min Scheduling）、AFCFS（Adaptive First Come First Service）、Fair的调度算法对比表明,Weather Cloud的调度算法不但减少了计算的等待时间,而且增加了完成的指令数量。     

Multihybrid job scheduling for fault-tolerant distributed computing in policy-constrained resource networks

Unpredictable fluctuations in resource availability often lead to rescheduling decisions that sacrifice a success rate of job completion in batch job scheduling. To overcome this limitation, we consider the problem of assigning a set of sequential batch jobs with demands to a set of resources with constraints such as heterogeneous rescheduling policies and capabilities. The ultimate goal is to find an optimal allocation such that performance benefits in terms of makespan and utilization are maximized according to the principle of Pareto optimality, while maintaining the job failure rate close to an acceptably low bound. To this end, we formulate a multihybrid policy decision problem (MPDP) on the primary-backup fault tolerance model and theoretically show its NP-completeness. The main contribution is to prove that our multihybrid job scheduling (MJS) scheme confidently guarantees the fault-tolerant performance by adaptively combining jobs and resources with different rescheduling policies in MPDP. Furthermore, we demonstrate that the proposed MJS scheme outperforms the five rescheduling heuristics in solution quality, searching adaptability and time efficiency by conducting a set of extensive simulations under various scheduling conditions.     

基于测距的无线传感器网络均衡式招募调度算法

节点调度是均衡无线传感器网络能量有效方法之一．分析基于测距的睡眠调度算法（RBSS）发现其招募节点能耗过大,造成其过早死亡,影响网络的生命周期．针对这个问题,本文在正六边形覆盖模型的基础上,基于能量均衡思想,提出基于测距的均衡式招募调度算法（RBDRS）．RBDRS算法将协作节点招募的任务转移到新招募的协作节点上,均衡网络能耗．招募节点通过测距招募距其最远的邻居节点作为协作节点,协作节点再依次为招募节点招募新的协作节点,直至无法招募到新的协作节点．仿真实验结果表明,与RBSS算法相比,在不增加额外开销的条件下,RBDRS算法能够有效减少工作节点数目,提高网络覆盖率,均衡网络能耗,延长网络生命周期.     

网络集群计算系统中的并行任务调度

基于多处理机并行任务调度模型，探讨网络集群计算系统中的并行任务调度问题，首先证明了一般网络集群计算系统中调度算法的可近似性难度，然后提出了三种不同的启发式算法：最大长度优先调度算法、最大宽度优先调度算法和最大面积优先调度算法；然后根据大量的模拟实验对这些算法以及文献中已提出的调度算法进行了比较分析，结果表明该文的启发式算法比文献中的算法在性能上效果更好。     

利用结构化对等网络构造计算资源共享平台

设计和构建了一个基于结构化对等网络的计算资源共享平台DHT-CRSP。它可以把因特网上用户提交的科学计算作业高效地映射到平台中合适的工作节点上运行,通过容错和安全机制,能保证系统的可靠性和正确性。描述了DHT-CRSP中支持的两种分布式哈希表：Chord协议节点树和CAN协议空间区域;分析了DHT-CRSP中高效的资源匹配算法。通过构建评测环境,运行各种负载与作业场景下的结果表明,DHT-CRSP系统可以获得好的负载均衡性能、低的资源匹配代价,它提供了一种构建高性能的桌面网格平台的新思路。     

Adaptive Task Pools: Efficiently Balancing Large Number of Tasks on Shared-address Spaces

Task based approaches with dynamic load balancing are well suited to exploit parallelism in irregular applications. For such applications, the execution time of tasks can often not be predicted due to input dependencies. Therefore, a static task assignment to execution resources usually does not lead to the best performance. Moreover, a dynamic load balancing is also beneficial for heterogeneous execution environments. In this article a new adaptive data structure is proposed for storing and balancing a large number of tasks, allowing an efficient and flexible task management. Dynamically adjusted blocks of tasks can be moved between execution resources, enabling an efficient load balancing with low overhead, which is independent of the actual number of tasks stored. We have integrated the new approach into a runtime system for the execution of task-based applications for shared address spaces. Runtime experiments with several irregular applications with different execution schemes show that the new adaptive runtime system leads to good performance also in such situations where other approaches fail to achieve comparable results.     

On construction of a well-balanced allocation strategy for heterogeneous multi-cluster computing environments

With the rapid increment of the heterogeneity of hardware devices, cluster computing has to encounter the problem of handling heterogeneous resources for exploiting the utilization of system resources. This paper introduces a new job allocation strategy based on multi-clusters in diskless environments. By adopting Ganglia as the resource monitor and Condor as the queue system, a heterogeneous multi-cluster system is also constructed with and without storage devices for evaluating the system performance. The proposed algorithm is called the Well-Balanced Allocation Strategy (WBAS) in which the scheduler dispatches MPI-based jobs to appropriate resources across multi-clusters. The strategy focuses on dispatching jobs to nodes with similar performance, thus equalizing execution times among all the required nodes. The WBAS is implemented on the constructed heterogeneous multi-cluster system to evaluate the performance of the scheduling strategy. The experimental results show that the proposed strategy performs well and could efficiently improve the system performance.     

基于资源域的分级调度模型

作业调度是要实现作业和资源的最佳匹配,协调对资源的征用矛盾.软件环境和硬件资源对作业运行都有影响,实验证明硬件资源对作业的影响很大,据此提出了基于资源域的分级调度模型,提供全局和局部两种调度方式.节点根据配置划分成不同的资源域,调度时根据作业类型匹配最佳的资源域,该模型有效利用了集群资源,减轻了Server的负担,并能提高系统的可扩展性.在全局调度中,还提供了预约机制调度高优先级作业.     

Min_c: Heterogeneous concentration policy for energy-aware scheduling of jobs with resource contention

In this paper, we address energy-aware online scheduling of jobs with resource contention. We propose an optimization model and present new approach to resource allocation with job concentration taking into account types of applications and heterogeneous workloads that could include CPU-intensive, diskintensive, I/O-intensive, memory-intensive, network-intensive, and other applications. When jobs of one type are allocated to the same resource, they may create a bottleneck and resource contention either in CPU, memory, disk or network. It may result in degradation of the system performance and increasing energy consumption. We focus on energy characteristics of applications, and show that an intelligent allocation strategy can further improve energy consumption compared with traditional approaches. We propose heterogeneous job consolidation algorithms and validate them by conducting a performance evaluation study using the Cloud Sim toolkit under different scenarios and real data. We analyze several scheduling algorithms depending on the type and amount of information they require.     

基于优先级和优化完成时间的网格调度算法

网格由大量的异构资源组成,具有复杂性、动态性和自治性特点。高效的网格调度算法可以充分利用网格系统资源,提高网格处理应用程序的能力。Min min算法是一个简单、快速、有效的调度算法,但由于总是先分配小任务而不能确保负载平衡。文中首先对网格系统中任务的数据传输和执行进行分析,计算并优化Min min算法的任务完成时间,再根据任务需求赋予任务优先级,通过优先级安排任务调度,提高算法负载平衡能力,最后在上述分析基础上提出POTE Min min(Priority and Overlap Transmission and Execution Min min)调度算法。     

KOALA: a co‐allocating grid scheduler

In multicluster systems, and more generally in grids, jobs may require co‐allocation, that is, the simultaneous or coordinated access of single applications to resources of possibly multiple types in multiple locations managed by different resource managers. Co‐allocation presents new challenges to resource management in grids, such as locating sufficient resources in geographically distributed sites, allocating and managing resources in multiple, possibly heterogeneous sites for single applications, and coordinating the execution of single jobs at multiple sites. Moreover, as single jobs now may have to rely on multiple resource managers, co‐allocation introduces reliability problems. In this paper, we present the design and implementation of a co‐allocating grid scheduler named KOALA that meets these co‐allocation challenges. In addition, we report on the results of an analysis of the performance in our multicluster testbed of the co‐allocation policies built into KOALA . We also include the results of a performance and reliability test of KOALA while our testbed was unstable. Copyright © 2007 John Wiley & Sons, Ltd.