| Summary: | 碩士 === 國立交通大學 === 資訊工程系 === 87 === Recently, the video server plays an important role in VOD or near VOD system. In general, a good video server can not only support playback operation, but also provide reliable service. In this thesis, we will propose an effective data placement scheme for supporting fault-tolerance in distributed video server environment, called Two Level Clustering (TLC) method. This method includes three parts, popularity-based replication, two level clustering, and shift load scheduling. Using popularity-based replication to calculate the number of replicas, we place the videos with two level clustering, and schedule with shift load scheduling policy. This method can support fault-tolerance ability and balance the load of the disks, thereby helping upgrade system performance crucial to archiving maximal server throughput and improve the system reliability. According to our preliminary performance evaluations, we have found that our method is superior than other schemes. The detailed information about design principles and performance evaluations will be described in the literature.
|
|---|
