MogileFS

From Wikipedia, the free encyclopedia

This article may not meet the general notability guideline or one of the following specific guidelines for inclusion on Wikipedia: Biographies, Books, Companies, Fiction, Music, Neologisms, Numbers, Web content, or several proposals for new guidelines. If you are familiar with the subject matter, please expand or rewrite the article to establish its notability. The best way to address this concern is to reference published, third-party sources about the subject. If notability cannot be established, the article is more likely to be considered for redirection, merge or ultimately deletion, per Wikipedia:Guide to deletion.
This article has been tagged since January 2008.

This article or section needs sources or references that appear in reliable, third-party publications.
Primary sources and sources affiliated with the subject of the article are generally not sufficient for a Wikipedia article. Please include more appropriate citations from reliable sources, or discuss the issue on the talk page.
This article has been tagged since June 2007.

MogileFS was created by Danga Interactive for use^[clarify] with LiveJournal. It is similar to the Andrew File System in its design goal. It is different from a traditional filesystem in that the user has to access files via an API. However, it is possible to implement the file system in user space using FUSE or a similar package.

MogileFS enables horizontal scaling of storage across any number of machines. Files are replicated between machines according to customizable rules such that at any time a number of drives or machines could be lost without data becoming unavailable. It is designed for high volume applications, such as high traffic websites, to spread storage across cheaper machines without relying on technologies such as NFS.

A typical MogileFS cloud consists of a database, some number of trackers, and some number of storage units.

The name is an anagram of "OMG files".

At least one startup uses MogileFS as a filestore for Hadoop Apache's distributed computing platform ^[1]. They chose this over Hadoop's built in file system, HDFS, as it was felt to be able to hold many small files better than HDFS, that only scales if files can be broken into blocks of 64MB or larger.