公开(公告)号：KR1020030051930A

公开(公告)日：2003-06-26

申请号：KR1020010081571

申请日：2001-12-20

Applicant: 한국전자통신연구원

Inventor： 김진미 ,  김학영

IPC: G06F15/16

Abstract: PURPOSE: A device and method for implementing a high availability of a cluster system is provided to improve a processing performance by restarting an application program having a trouble or distributing a task of a server having a trouble to other system of a cluster in the case that a trouble is generated in one system of the cluster or an operation of the application program. CONSTITUTION: A plurality of clients(100-1, ... , 100-n) are connected to a cluster(200) and receive a high availability web service. A main control server(210) loads a management object for a high availability management for monitoring the clients(100-1, ... , 100-n), detecting an error, and starting/stopping an error recovery, and the clients(100-1, ... , 100-n) select and connect to a proper node out of various group nodes in the cluster(200). In the case that a web service which is operated by the main control server(210) is stopped caused by a trouble, a sub control server(220) is connected to the main control server(210), and executes a recovery of the web service when an error is detected. If the error is recovered, the sub control server(220) copies web service related information, and executes a high availability web service. A plurality of main nodes(230-1-1, ... ,230-n-n) supply a service such as a web server and a file transfer protocol server to the clients(100-1, ... , 100-n) through a network and the Internet. If a service is stopped caused by a trouble while supplying a web service by the main nodes(230-1-1, ... ,230-n-n), a plurality of sub nodes(231,232,233) maintain the web service continuously. Disk mirroring(215,225) store web service related information for copying the web service related information in an error recovery.

Abstract translation: 目的：提供一种用于实现集群系统的高可用性的设备和方法，以通过在具有故障的应用程序或将具有故障的服务器的任务分发给集群的其他系统的情况下重新启动应用程序来提高处理性能， 在集群的一个系统或应用程序的操作中产生故障。 构成：将多个客户端（100-1，...，100-n）连接到集群（200）并接收高可用性web服务。 主控制服务器（210）加载用于高可用性管理的管理对象，用于监视客户端（100-1，...，100-n），检测错误和启动/停止错误恢复以及客户端 100-1，...，100-n）从集群中的各个组节点（200）中选择并连接到正确的节点。 在由主控制服务器（210）操作的网络服务由于故障而停止的情况下，副控制服务器（220）连接到主控制服务器（210），并执行网络的恢复 检测到错误时的服务。 如果错误被恢复，则子控制服务器（220）复制web服务相关信息，并且执行高可用性web服务。 多个主节点（230-1-1，...，230-nn）向客户机（100-1，...，100-n）提供诸如web服务器和文件传输协议服务器的服务， 通过网络和互联网。 如果在由主节点（230-1-1，...，230-n-n）提供Web服务的情况下由于故障而停止服务，则多个子节点（231,232,233）连续地维护web服务。 磁盘镜像（215,225）存储用于在错误恢复中复制Web服务相关信息的Web服务相关信息。

公开(公告)号：KR1019980036297A

公开(公告)日：1998-08-05

申请号：KR1019960054855

申请日：1996-11-18

Applicant: 한국전자통신연구원

Inventor： 김진미 ,  표창우 ,  한경숙

IPC: G06F9/50 ,  G06F9/06

Abstract: 본 발명은 필요한 레지스터 수를 Chaitin 방식보다 작게 하며, 복사문에 의한 변수의 특정 레지스터 선호, 변수 사용 위치에 의한 변수의 특수 기능 레지스터 선호를 단순하며 통일된 체계에서 처리하고, 프로그램 구조를 이용한 레지스터 할당에 있어서도 그래프 분할 정보를 전역적으로 활용하게 하여 프로그램 구조에 따라 레지스터 할당이 진행될 때 전반적인 간섭 관계를 쉽게 파악할 수 있는 그래프 분할에 의한 레지스터 할당 방법에 관해 개시된다.