A Fast File Deletion Method in Distributed File System

Abstract

The invention provides a method for fast deleting files in a distributed file system. According to the method, when the files in the distributed file system are deleted, corresponding metadata information on a metadata server is added to a recycle catalog, a deleting success signal is returned, the metadata server scans the recycle catalog at regular time and renames data objects corresponding to the files with new object names, the object names are file objects, to be distributed to new files, in the system in a next batch, in this way, when new file data are written in, renamed objects of the newly recycled files can be directly used, and when the files are read, the size of the effective data is judged according to the size of the files recorded in the metadata, so that the purpose of fast deleting the files without really deleting the disc objects or affecting normal read and write is achieved. Under most circumstances, it is unnecessary to delete the file data disc objects, and the files are really fast deleted without generating disc data IO or affecting read and write IO of the normal data.

Description

technical field [0001] The invention relates to a deletion method, in particular to a fast file deletion method in a distributed file system. Background technique [0002] A distributed file system generally includes a client, a metadata server and a data server. The client is responsible for formulating the access interface of the file data, the metadata server handles the layout and attributes of the file, and the data server stores the data content of the file. [0003] For a distributed file system, the ability to store massive amounts of data is its most important feature. When a large number of files are stored in the system and the files need to be deleted in large quantities, a large amount of disk io will often be generated, which directly affects normal data reading and writing. , reducing the read and write performance of the entire system. [0004] The traditional method is to hide the metadata of the file, and then recover the corresponding data files in the me...

AI Technical Summary

Problems solved by technology

[0003] For a distributed file system, the ability to store massive amounts of data is its most important feature. When a large number of files are stored in the system and the files need to be deleted in large quantities, a large amount of disk io will often be generated, which directly affects normal data reading and writing. , reducing the read and write performance of the entire system

[0004] The traditional method is to hide the metadata of the file, and then recycle the corresponding data files in the metadata in the background. In order to reduce the impact on normal reading and writing, we can only limit the number of files recycled each time, although this can reduce the impact on the normal operation of the system. The impact of reading and writing, but greatly reduces the speed of deletion, that is, the speed at which disk space is re-used

Images