Exploring the allocation of heterogeneous file systems in cloud storage systems

• Main Authors: Wei-Peng Chen, 陳煒鵬
• Other Authors: Chuan-Ming Liu
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/y76476

碩士 === 國立臺北科技大學 === 資訊工程系研究所 === 104 === The micro-enterprises and small-scale enterprises are raised in recent decade. They build a considerable number of information systems, resulting in a large number of operations, as well as a large amount of data stored. In the human resources and budget conscious premise, internal IT unit must have the ability to handle large data, and make proper data backup. To solve the huge demand for storage and cost, cloud storage services have become one of the important solutions. These solutions include first storage resource virtualization, the construction of storage pools, the supplies of online data access and the backup of web services. In this study, we compare the time cost and the efficiency with three common file systems, namely GlusterFS, Lustre, and Hadoop. Experimental results show as follows. First, the Hadoop has the highest performance when the storage is larger than 200MB file in these three systems. Second, the Lustre has the highest performance when the storage is less than 200MB and more than 4MB file. Final, the GlusterFS has the highest performance when the storage is less than 4MB file.