International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395 -0056
Volume: 04 Issue: 03 | Mar -2017
p-ISSN: 2395-0072
www.irjet.net
An Efficient Approach to Manage Small Files in Distributed File Systems Aakash Patil, Ganesh Sagare, Kunal Saraf (BE in Computer Engineering, Sandip Institute of Engineering and Management, Nashik.) Prof. Sujit. A. Ahirrao Assistant Professor, Department of Computer Engineering ---------------------------------------------------------------------***--------------------------------------------------------------------Abstract: Nowadays, to manage excessive number of small files is became a challenge in Distributed File System. Currently, the combined block storage technique is used to store the files this technique is used in existing system such as Extfs and Xfs. This technique is liable to inefficiency when accessing files randomly. We present the proposed system to manage small files which is based on simple metadata and storage architecture. Our system focuses on replacing the existing system drawbacks in Data servers that used to store excessive number of small files and retrieval of files in a better way. We designed new metadata structure which will decrease the size of original metadata that will help to increase the speed of file accessing. Keywords: Information System , Information Storage And Retrieval. Indexing Methods, Content Analysis Computing Methodologies, Documents Processing, Various types of files. 1. Introduction: We know that Metadata consist of data related data that means in file system metadata contains the information which is helpful to search the files in file systems for eg. Address of the file, size of the file, modified date of updated information etc. Nowadays, Everyone is using social networking and e-commerce websites for communication and purchasing purpose by considering the usage of the websites which required to store the data which is small in size then there is the difficulty in storing and retrieving the files which are smaller in size and the number of this files are bulk because of many users are frequently uploading or modifying the data in the storage space. So, the managing this small files is became a problem in distributed file system because of the metadata generated by the files is bigger in size. In some cases the files are rarely modified or updated and the size of this file is in between 1kb’s to 10kb’s such as pictures, text etc. uploaded on social networking and e-commerce websites in daily or timely basis. Distributed file system is based on storing and accessing files based on simple client-server architecture. In distributed file system all data is copied and placed on the different data servers and the information about the data is stored in which are then connected in network. A client or user searches the file using metadata server other than the using the actual location of that file the same process is used in existing system, client request the file which is stored in a distributed file system by using two phases. 1.Client sends the query containing about the data needed to the metadata server and gets the IP address of data server which stores the target file. 2.In next phase connection between data server and user is established and granted for fetching the data file. Why we are shrinking the size of metadata ? In our proposed system the main reason behind shrinking the size of metadata is, in DFS when we are storing the file, the size of its metadata is big in size because of it contains every attributes as discussed earlier. Because of these the accessing speed of a particular file takes more time. In our system the metadata will contain only two things that are size of the file and physical address of that file so that accessing speed can be increased.
Š 2017, IRJET
|
Impact Factor value: 5.181
|
ISO 9001:2008 Certified Journal
|
Page 2311