Technology news and Jobs 
VIRTUALISATION Enter ext4, the filesystem of the future
VIRTUALISATION Enter ext4, the filesystem of the future | Enter ext4, the filesystem of the future |
|
| by Sam Varghese | |
| Wednesday, 24 June 2009 | |
|
Page 2 of 3 Chris Samuel: The downsides of ext3 has traditionally been that doesn't handle very heavy workloads well (you can often get lots of processes waiting to do writes) and has a limited filesystem size on common architectures (8TB is a typical limit). Featured Whitepaper
5 Best Practices for Smartphone Support
The latest kernel release (2.6.30) alters the default journalling options to improve performance and reduce latency (the time taken to do file operations) whilst trying to reduce the impact on badly written applications. ext3 is also what is known as an indirect block based filesystem, where files are made up of discrete blocks on disk and there is a structure of inodes that eventually lead you to where the data is stored. While this is fine for smaller files when you start storing large data files it can cause significant inefficiencies as you have to traverse all this information to be able to read back the entire file. iTWire: I've heard mention that XFS has some advantages over ext3. Chris Samuel: XFS is not a block based filesystem like ext3, it uses what are called "extents" instead. These are contiguous chunks of disk that can dramatically improve efficiency and performance when dealing with large files. It also can have far larger partition sizes than ext3 is capable of (up to 8EB) as well as supporting real time functionality. iTWire: Fedora has started using ext4 as the default filesystem. What are some of the pros and cons of such a decision? Chris Samuel: ext4 is a much bigger change from ext3 than ext3 was in turn from ext2. Ext3 essentially just added journalling to ext2, whereas ext4 moves to an extent based filesystem with other features such as delayed allocations (like XFS) to allow the allocator to be more intelligent about how it lays things out on disk and much bigger filesystem sizes (though the programs to create ext4 filesystems can't actually make them for you yet). Another nice feature is the fact that the journal data is checksummed so the filesystem can spot any corruption after a crash. So the upsides are all the new features, but the downside is that there is an awful lot of new code here and while it has been in the kernel for a while marked "experimental" it's not had the years of testing and hardening that ext3 has gone through. That said I've been running my home filesystem on ext4 since the very end of 2009 and haven't had any problems with it (yet!). iTWire: When do you see ext4 as becoming the norm for GNU/Linux distributions? Chris Samuel: I think you're going to see most distros swapping to it over the next 12 months. For instance the second alpha test release of the next Ubuntu distribution is using it by default, as is the second milestone release for OpenSUSE 11.2 (due out later this year). CONTINUED |
| < Next story in category | Previous story in the category > |
|---|
