JOURNALLED FILE SYSTEM

5/5/2001

Journalled File System: File systems that keep track of transactions and install or remove entire transaction(s) in the reverse order in which they took place. The intent is to prevent situations where some of a set of related files are updated and others are not. A second goal is the ability to back off a file system to a prior state in a straightforward manner. A third is the desire to avoid the need for time consuming file system quality checks such as scandisk (msdos/windows-FAT) or fsck(unix-EXT2) on boot. Journalled File systems are an import from the world of Mainframes and thus are found in the PC world primarily in Unix/Linux systems. NTFS is the only journalled file system in the Microsoft world. NTFS's journalling is apparently somewhat limited compared to more elaborate systems.

Journalled recovery is similar to the recovery techniques used by data bases to recover to a consistent state after a crash. For the most part, this gets around the need to run file by file quality checks on the file system. Those checks can take unreasonable amounts of times for large file systems.

Journalling is said to have little impact on performance. Journalling is generally built in as one feature of a complex set of features. Since it is not a discrete add-on that can simply be disabled, it is not easily possible to determine what the performance impact of journalling itself really is.

The two best known Linux journalled file systems are ReiserFS and XFS. Other Journalled file systems include JFS, Ext3 and (maybe) Tux2. ReiserFS is supported by Linux Kernels.

https://web.archive.org/web/20100612172755/http://linuxgazette.net/issue55/florido.html