| Summary: | 碩士 === 國防大學理工學院 === 資訊科學碩士班 === 99 === This thesis aims to design a cloud storage environment for the military Learning Management System. All server nodes in our cloud environment are cooperating with one another in a decentralized approach to transmit file blocks. We propose the Dynamic Assignment Fast Replica algorithm (DAFR) to transmit content files by cutting each content file into several blocks in accordance with network transmission capacity between nodes. Thus, the DAFR algorithm can make faster these file blocks copied into all nodes in the environment.
Moreover, we design a program to simulate the file transmission. Consequently, a comparison is made between Fast Replica algorithm, proposed by J. Amugharaj, and our Dynamic Assignment Fast Replica algorithm. The demonstration shows that the file transmission with our DAFR algorithm in a cloud storage environment is faster than Fast Replica algorithm.
To verify the applicability of our algorithm when applied in an extreme low and different transmission capacity among nodes in the cloud environment, Clustered Fast Replica (CFR) is proposed, and a test has been carried out to examine the file transmission performance. The result shows that Fast Replica algorithm is not suitable for this network environment and the performance is not what we expected.
Additionally, this thesis also proposes a load balanced approach in the Learning Management System. When user obtains materials from the cloud storage environment, it can decentralize the load and then reduce the time of downloading.
|
|---|
