• DocumentCode
    3772422
  • Title

    File System Allocation in Cloud Storage Services with GlusterFS and Lustre

  • Author

    Wei-Chih Huang;Chuan-Chi Lai;Chun-An Lin;Chuan-Ming Liu

  • Author_Institution
    Dept. of Comput. Sci. &
  • fYear
    2015
  • Firstpage
    1167
  • Lastpage
    1170
  • Abstract
    Nowadays, the information systems for enterprises become popular and these systems produce an extreme huge amount of data. The produced data are all important and should be kept for further analysis. However, as this trend continues, the lack of human resources and budgets makes the ways to efficiently store and analyze data challenging. To solve this predicament, Cloud Storage Service (CSS) provides one of the possible solutions. For CSS, the storage resources become virtualized and can form a storage pool. Then, using the Internet, CSS allows the data can be accessed and backup as on the local storage. This study explores appropriate approaches for building the file systems with lower-cost to provide highly reliable and efficient CSS for enterprises. Two storage file systems are considered: GlusterFS and Lustre. GlusterFS is widely used in distributed environments while Lustre is mainly for high-performance computing. With a series of comparisons and analyses, we can find the appropriate file systems for different applications with better utilization and high performance for accessing files. The small capital enterprise companies thus can build their own CSSs without abundant resources. As a result, Lustre Stripe type is better and efficient for write-in under the low-cost circumstances. On the other hand, Lustre Distributed type can read files fluently and excellently.
  • Keywords
    "File systems","Cloud computing","Servers","Conferences","Performance evaluation","Metadata"
  • Publisher
    ieee
  • Conference_Titel
    Smart City/SocialCom/SustainCom (SmartCity), 2015 IEEE International Conference on
  • Type

    conf

  • DOI
    10.1109/SmartCity.2015.230
  • Filename
    7463885