Reducing rename overhead in full-path-indexed file system

0Citations
Citations of this article
2Readers
Mendeley users who have this article in their library.
Get full text

Abstract

Full-path-indexed file systems use a key-value database to store the full path names of files and their metadata. With this pattern, the I/O efficiency can be improved because data is placed on persistent storage in scan order. However, it introduces intolerable overhead on renaming a directory because of the modification on the full path names of files under that directory. In this paper, we introduce prefix replacement mechanism on B+-tree to accelerate renaming directories on full-path-indexed file systems. It consists of three steps: pre-scan prefix deletion, key replacement and floating-split bulk insertion. Unnecessary searches and compares are reduced in these mechanisms. We use Kyoto Cabinet as the key-value database, and implement prefix replacement mechanism on it. We run tests on two benchmarks, the first is generated by Mdtest [18], and the second is the source code of Linux [19]. Compared with LocoFS [4], one kind of full-path-indexed file system, our design is about 5× faster to rename large directories, and the performance is basically same on small directories.

Cite

CITATION STYLE

APA

Wang, L., Lu, Y., Li, S., Yang, F., & Shu, J. (2019). Reducing rename overhead in full-path-indexed file system. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11719 LNCS, pp. 43–54). Springer Verlag. https://doi.org/10.1007/978-3-030-29611-7_4

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free