Correlation-Merge: A Small Files Performance Optimization for SWIFT Object Storage

Xiguang Wang,Xueming Qiao,Weiyi Zhu,Yang Gao,Dongjie Zhu,Zhongchuan Fu
DOI: https://doi.org/10.1007/978-3-319-49568-2_32
2017-01-01
Abstract:In recent years as the internet applications e.g. e-business, web browsing, video website, and social network etc. prevail, data increase at an unprecedented rate. Cloud computing plays an significant role in the big-data era. However, most cloud storage platforms, e.g. SWIFT of OpenStack, aims at optimizing large files performance. Small files with average size less than 15 KB severely hurdles the storage performance. In this paper a framework, namely Correlation-Merge, is proposed to optimize the storage performance of small files in cloud. An optimization layer, namely SFAL (small files additional layer) between Object Server and file system is introduced. Experimental results show that the optimization approach acquires satisfactory efficacy.
What problem does this paper attempt to address?