• 정보과학회 논문지
• /
• 제42권1호
• /
• pp.15-22
• /
• 2015

본 논문에서는 대규모의 계산 작업을 고성능으로 처리해야 하는 신약재창출 시뮬레이션 분야에 분산 슈퍼컴퓨팅 기술을 적용한 사례에 대해 논의하고자 한다. 신약재창출이란 기존에 알려진 약물의 새로운 적응증을 규명하는 것을 의미하며, 이러한 신약재창출은 비교적 짧은 수행시간을 갖는 대규모의 도킹(docking) 연산들을 고성능으로 처리해야한다는 점에서 Many-Task Computing (MTC) 성격을 지니고 있다. 이러한 MTC 응용들의 대표 사례로서 신약재창출 시뮬레이션을 분산 슈퍼컴퓨팅 환경 기반의 HTCaaS 시스템에 적용하였으며, 이를 통해 효율적인 작업 배포, 동적인 자원 할당 및 로드 밸런싱, 안정성 및 다양한 자원들의 효율적인 통합 등이 이러한 과학 응용들을 지원하는 데 있어 필수적인 기능임을 확인할 수 있었다.

• 한국정보과학회논문지:시스템및이론
• /
• 제27권6호
• /
• pp.549-557
• /
• 2000

최근 응용 프로그램들은 복잡한 데이타로 구성되어 있기 때문에 이를 효율적으로 처리할 수 있는 분산 메모리 기계(Distributed Memory Machine : DMM)의 필요성이 대두되었다. 특히 태스크 스케줄은 태스크 사이의 통신 시간을 최소화하여 응용 프로그램 전체의 실행 시간을 단축시키는 기법으로서, DMM의 성능을 향상시키는 매우 중요한 요소이다. 기존의 태스크 중복 스케줄(Task Duplicated based Scheduling : TDS) 기법은 두 개의 태스크 사이에 통신 시간이 많이 소요되는 것들을 하나의 클러스터(cluster)로 스케줄함으로써 통신 시간을 단축하여 실행 시간을 향상시키는 기법이다. 그러나 데이타를 전달하는 태스크와 이 태스크로 데이타를 전달받는 태스크 사이의 통신 시간을 최적화 하지 못하는 단점을 가진다. 따라서 본 논문에서는 이 두 태스크 사이의 최적화에 근접한 통신 시간을 갖는 개선된 중복 스케줄 (Modified Task Duplicated based Scheduling : MTDS) 기법을 제안하였다. 이 기법은 데이타를 전달한 태스크들을 클러스터링하기 위해 데이타를 전달받은 태스크에서 최적화 조건을 적용하여 검사한다. 그 결과 태스크 사이의 통신 시간을 단축하여 전체 태스크 실행 시간을 최소화하였다. 또한 시스템의 모델링을 통하여 MTDS 기법이 최상의 경우 TDS 기법보다 태스크 실행 시간을 70% 단축 시켰고 최악의 경우 TDS 기법과 동일한 실행 시간을 얻으므로 제안된 기법이 기존의 기법보다 우수함을 입증하였다.

• 제어로봇시스템학회논문지
• /
• 제3권5호
• /
• pp.542-554
• /
• 1997

In this paper, we present a resource conscious approach to designing distributed real-time systems as an extension of our original approach [8][9] which was limited to single processor systems. Starting from a given task graph and a set of end-to-end constraints, we automatically generate task attributes (e.g., periods and deadlines) such that (i) the task set is schedulable, and (ii) the end-to-end timing constraints are satisfied. The method works by first transforming the end-to-end timing constraints into a set of intermediate constraints on task attributes, and then solving the intermediate constraints. The complexity of constraint solving is tackled by reducing the problem into relatively tractable parts, and then solving each sub-problem using heuristics to enhance schedulability. In this paper, we build on our single processor solution and show how it can be extended for distributed systems. The extension to distributed systems reveals many interesting sub-problems, solutions to which are presented in this paper. The main challenges arise from end-to-end propagation delay constraints, and therefore this paper focuses on our solutions for such constraints. We begin with extending our communication scheme to provide tight delay bounds across a network, while hiding the low-level details of network communication. We also develop an algorithm to decompose end-to-end bounds into local bounds on each processor of making extensive use of relative load on each processor. This results in significant decoupling of constraints on each processor, without losing its capability to find a schedulable solution. Finally, we show, how each of these parts fit into our overall methodology, using our previous results for single processor systems.

• 한국군사과학기술학회지
• /
• 제17권2호
• /
• pp.167-174
• /
• 2014

In this paper, we address a task allocation problem for a robot team that performs a search and attack mission. The robots are limited in sensing and communication capabilities, and carry different types of resources that are used to attack a target. The environment is uncertain and dynamic where no prior information about targets is given and dynamic events unpredictably happen. The goal of robot team is to collect total utilities as much as possible by destroying targets in a mission horizon. To solve the problem, we propose a distributed task allocation framework incorporated with effective resource management based on resource welfare. The framework we propose enables the robot team to retain more robots available by balancing resources among robots, and respond smoothly to dynamic events, which results in system performance improvement.

• Journal of Electrical Engineering and Technology
• /
• 제11권4호
• /
• pp.987-992
• /
• 2016

This study introduces an example environment where wireless devices are mobile, devices use dynamic voltage scaling, devices and tasks are heterogeneous, tasks have deadline, and the computation and communication power is dynamically changed for energy saving. For this type of environment, the efficient system-level energy management and resource management for task completion can be an essential part of the operation and design of such systems. Therefore, the resources are assigned to tasks and the tasks may be scheduled to maximize a goal which is to minimize energy usage while trying to complete as many tasks as possible by their deadlines. This paper also introduces mobility of nodes and variable transmission power for communication which complicates the resource management/task scheduling problem further.

• Journal of Computing Science and Engineering
• /
• 제10권1호
• /
• pp.9-20
• /
• 2016

Several task clustering heuristics are proposed for allocating tasks in heterogeneous systems to achieve a good response time in data intensive jobs. However, one of the challenging problems is the process in task scheduling after task allocation by task clustering. We propose a task scheduling method after task clustering, leveraging worst schedule length (WSL) as an upper bound of the schedule length. In our proposed method, a task in a WSL sequence is scheduled preferentially to make the WSL smaller. Experimental results by simulation show that the response time is improved in several task clustering heuristics. In particular, our proposed scheduling method with the task clustering outperforms conventional list-based task scheduling methods.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제10권1호
• /
• pp.117-135
• /
• 2016

Performance enhancement is one of the most important issues in high performance distributed computing systems. In such computing systems, online users submit their jobs anytime and anywhere to a set of dynamic resources. Jobs arrival and processes execution times are stochastic. The performance of a distributed computing system can be improved by using an effective load balancing strategy to redistribute the user tasks among computing resources for efficient utilization. This paper presents a multi-class load balancing strategy that balances different classes of user tasks on multiple heterogeneous computing nodes to minimize the per-class mean response time. For a wide range of system parameters, the performance of the proposed multi-class load balancing strategy is compared with that of the random distribution load balancing, and uniform distribution load balancing strategies using simulation. The results show that, the proposed strategy outperforms the other two studied strategies in terms of average task response time, and average computing nodes utilization.

• ETRI Journal
• /
• 제20권4호
• /
• pp.346-360
• /
• 1998

It is challenging to parallelize problems with irregular computation and communication. In this paper, we propose an asynchronous algorithm for balancing unpredictable workload on distributed-memory machines. By using an initial workload estimate, we first partition the computations such that the workload is distributed evenly across the processors. In addition, we perform task migrations dynamically for adapting to the evolving workload. To demonstrate the usefulness of our load balancing strategy, we conducted experiments on an IBM SP2 and a Cray T3D. Experimental results show that our task migration strategy can balance unpredictable workload with little overhead. Our code using C and MPI is portable onto other distributed-memory machines.

• 한국멀티미디어학회논문지
• /
• 제7권5호
• /
• pp.755-765
• /
• 2004

최근에 많은 유휴 상태의 호스트 자원들을 이용한 인터넷 기반 분산/병렬 컴퓨팅은 대용량 작업처리와 여러 중요 논제들에 대해 그 유용성이 증명되고 있다. 대용량 작업이 수행되는 동안, 작업에 참여하는 호스트의 성능과 상태 변화에 대처하기 위한 실시간 모니터링 기능이 요구된다. 본 연구에서는 글로벌 컴퓨팅 (global computing) 인트라스트럭처(infrastructure)로 구축된 인터넷 기반 분산/병렬 처리 프레임워크인 PDP(Parallel Distributed Processing)상의 실시간 모니터링 및 시각화에 대한 내용을 소개한다.

• TTA 저널
• /
• 통권164호
• /
• pp.39-44
• /
• 2016