클러스터형 VOD서버에서 고가용성을 고려한 자체 복구 시스템 = Self Recovery System With High Availability in Clustered VOD Server

다국어 초록 (Multilingual Abstract)

While clustered VOD (Video On Demand) servers that have been proposed for building large-scale VOD system recently, have the advantages of high availability and scalability, the system level reliability of the clustered VOD servers could be decreased by adding more servers to scale up the capacity of the server. To alleviate this reliability problem and support server-level fault tolerance, this paper proposes and analyzes self recovery system based on RAID-3 and 4 with pipelining concept. The proposed self recovery system could increase the efficiency of the disk by accessing the disk with large size block as in RAID-4. The network resource could be used efficiently and the workload of memory could be decreased by sending data in small size block as in RAID-3. By introducing pipeline concept, the recovery tasks are processed at several nodes in parallel so that the workload in CPU, network, and memory could be distributed to several nodes.