• Proceedings of the Korean Information Science Society Conference
• /
• 2001.10a
• /
• pp.73-75
• /
• 2001

다수의 단일노드를 고속의 네트웍으로 연결하여 단일 시스템처럼 인식할 수 있는 클러스터 DBMS 를 효율적으로 관리하기 위해서는 클러스터 관리 도구가 필요하나 관련 연구는 미흡한 실정이다. 본 논문에서는 클러스터 DBMS을 위한 관리도구로써 노드상태를 모니터링 할 수 있는 모니터링 도구를 설계한다. 아울러 이러한 모니터링 결과를 기반으로 각 노드의 부하를 측정하고 사용자의 요구를 적절히 분산시킴으로써 클러스터 DBMS의 성능을 향상을 도모할 수 있는 부하 분산 관리기를 설계한다.

• The Transactions of the Korea Information Processing Society
• /
• v.2 no.5
• /
• pp.768-776
• /
• 1995

In a distributed computer system with star topology, the processing power of central node may be greater than that of peripheral nodes and the job arrival rate of each node may be different. The performance of load balancing may be very different according to the selection of sender criteria and receiver criteria for job transfer. But the optimal decision of sender and receiver criteria is very difficult in heterogeneous distributed computer systems. In this paper we propose effective destination decision policies using dynamic criteria such as mean job response time of a node and estimated response time of a job, instead of fixed threshold for dynamic load balancing.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.04a
• /
• pp.748-750
• /
• 2004

AODV나 DSR과 같은 Reactive 라우팅 방식은 트래픽이 발생되는 시점에서 경로를 탐색하는 방법으로서 최적 메시지 전송 경로가 유효하다면 트래픽의 증가와 과도한 에너지 소비를 고려하지 않고 계속적인 동일 전송 경로의 사용으로 인해 특정 노드에 대하여 에너지 소모가 심화될 뿐만 아니라 전송지연이 발생된다. 본 논문에서는 과중한 트래픽 상태를 분산 시켜 전체 네트워크의 균형적인 에너지 소비를 위하여 Reactive 라우팅 방법에서 특정 노드의 트래픽 상태, 에너지 상태, 라우팅 시간을 고려한 노드 상태 기반 부하 분산 방법(NSLB： Node Status-based Load Balancing Method)을 제안한다.

• Journal of IKEEE
• /
• v.6 no.1 s.10
• /
• pp.87-93
• /
• 2002

In this paper, an advanced load balancing algorithm in nth order Hypercube distributed system has been proposed. The new algorithm uses centralized load-balancing to avoid blocking phenomenon and processor thrashing, and shows the results which makes loads to approach average value of loads. The new algorithm is compared with several other algorithm and it shows a merit in cost function value.

• 한국공간정보시스템학회:학술대회논문집
• /
• 2005.11a
• /
• pp.67-73
• /
• 2005

최근 이동체의 위치 데이타를 활용하는 위치 기반 서비스에 대한 관심이 급증하고 있다. 이러한 위치 기반 서비스에서 이용되는 대용량 위치 데이타를 효율적으로 관리하기 위한 아키텍처로서 클러스터 기반 분산 컴퓨팅 구조를 갖는 GALIS(Gracefully Aging Location Information System) 아키텍처가 제안되었다. GALIS는 비균등 2-단계 그리드를 사용하여 노드들의 부하 분산 및 색인을 수행한다. 하지만 비균등 2-단계 그리드의 분할 알고리즘은 이동체가 특정 지역에 편중되는 경우 불필요한 노드를 생성하는 문제를 가지고 있다. 따라서 본 논문에서는 이동체의 다양한 분포에 대하여 더욱 효율적인 노드 분할 알고리즘을 제시한다. 본 논문에서 제시한 노드 분할 알고리즘은 이동체의 현재 위치에 따른 공간적 분포를 고려하기 때문에 이동체가 특정 지역에 편중되는 경우에도 불필요한 노드를 생성하지 않고 효율적인 부하 분산을 수행할 수 있으며, 분산 시스템에서 중요시되는 균형 있는 부하 분산을 수행할 수 있다. 또한, 가상 노드 분할 시뮬레이터를 구현하여 다양한 이동체 데이타 분포 형태에 대해 실험하였으며, 이러한 실험을 통하여 기존의 알고리즘보다 더욱 효율적으로 노드를 분할하는 것을 검증하였다.

• The Journal of the Korea Contents Association
• /
• v.6 no.12
• /
• pp.136-146
• /
• 2006

In Wireless Sensor Network(WSN) a sensor node transfers sensing data to the base-node through multi-hop because of the limited transmission range. Also because of the limited energy of the sensor node, the sensor nodes are required to consume their energy evenly to prolong the lifetime of the network. LMPR is a routing protocol for WSN, LMPR configures the network autonomously based on level which is the depth from the base-node, and distributes the transmission and computation load of the network to each sensor node. This paper implements LMPR on TinyOS and experiments on the performance of LMPR in WSN. As the result, the average of the received rate of LMPR is 91.39% and LMPR distributes the load of the transmission and computation about 4.6 times compare to the shortest cost routing protocol. We expect LMPR evenly distributes the transmission and computation load of the network to each node, and the lifetime of the network will be longer than it used to be.

• Journal of Internet Computing and Services
• /
• v.11 no.3
• /
• pp.65-74
• /
• 2010

In the ubiquitous healthcare, automated diagnosis is commonly achieved by an agent system to provide intelligent decision support and fast diagnosis result. Mobile agent technology is used for efficient load distribution by migrating processes to a less loaded node which is considered in our design of a ubiquitous healthcare system. This paper presents a framework for ubiquitous healthcare technologies which mainly focuses on mobile agents that serve the on-demand processes of an automated diagnosis support system. Considering the efficient utilization of resources, a balanced clustering for the load distribution of processes within nodes is proposed. The proposed algorithm selects overloaded nodes to migrate processes to near nodes until the load variance of the system is minimized. Our proposed balanced clustering efficiently distributes processes to all nodes considering message overheads by performing the migration to the near nodes.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.9 no.5
• /
• pp.1142-1148
• /
• 2005

We proposes an algorithm of measuring computation loads to distribute workload of clients. The key of the algorithm is to transfer a suitable amount of processing demand from senders to receivers in relocating intrinsically. This amount is determined dynamically during sender-receiver negotiations. Factors considered when this amount is determined include processing speeds of different nodes, the current load state of both sender and receiver, and the processing demands of tasks eligible for relocation. We also propose a load state measurement scheme which is designed particularly for heterogeneous systems. Based on this analysis, we design new algorithm for supporting heterogeneous distributed web server system, and compare their performance against other existing algorithms. We show that the new algorithm improve the performance of CPU utilization and response time.

• Journal of Korea Multimedia Society
• /
• v.9 no.9
• /
• pp.1107-1117
• /
• 2006

In this paper, a load balancing method using ring network structure in the Grid database is proposed. In the Grid database, generally, data is replicated for performance and availability. And, user's request is transferred to node and processed in that node which has target data. But, in such environment, a decline of performance can be occurred because unbalanced workload. A traditional research is proposed to solve unbalanced load problem. However, the Grid database has a number of systems and user's request always changes dynamically. Therefore, a traditional research can not be applied. The proposed method connects each node which has a same replicated data through ing network structure. If workload is overflowed in some node, user's request is transferred to a linked node which has a target data. And, this node stops another request processing until workload is decreased. Then, it changes the link structure through sending a message to a previous node, to stop request forwarding from a previous node. This paper shows a proposed method increases performance than existing research through performance evaluation and is more suitable for a complex and dynamic environment.

• The Transactions of the Korea Information Processing Society
• /
• v.6 no.12
• /
• pp.3456-3468
• /
• 1999

Most of previous parallel join algorithms assume a database partition system(DPS), where each database partition is owned by a single processing node. While the DPS is novel in the sense that it can interconnect a large number of nodes and support a geographically distributed environment, it may suffer from poor facility for load balancing and system availability compared to the database sharing system(DSS). In this paper, we propose a dynamic load balancing strategy by exploiting the characteristics of the DSS, and then extend the conventional hash join algorithms to the DSS by using the dynamic load balancing strategy. With simulation studies under a wide variety of system configurations and database workloads, we analyze the effects of the dynamic load balancing strategy and differences in the performances of hash join algorithms in the DSS.