Disk Subsystem Enhancements
The 2.4 kernel continues to expand the general usefulness of Linux as a disk subsystem by increasing the list of exotic filesystem and partition types it understands. Examples include UDF (used by DVDs and CDRWs), NFS v3 client and server support, and partition support for almost anything you can find. On a more abstract note, devfs (/dev file system) support migrates /dev from being a real directory full of device files into a virtual filesystem, very similar to /proc. The upshot of this change is no more having to remember device numbers and mknod device files when you add devices to your system. Actually, MAKEDEV takes care of this for you in most cases, so the real advantage is performance.
Are hardware RAID devices a tad too pricey for your environment? The 2.4 kernel now offers boot support for software RAID devices in the "multiple-device support" section of the kernel. For those of you who have been yearning for the flexibility of repartitioning on the fly, LVM (Logical Volume Manager) support has arrived. The following description is taken from the kernel help file, which explains it as well as I could:
[The LVM] driver lets you combine several hard disks, hard disk partitions, multiple devices, or even loop devices (for evaluation purposes) into a volume group. Imagine a volume group as a kind of virtual disk. Logical volumes, which can be thought of as virtual partitions, can be created in the volume group. You can resize volume groups and logical volumes after creation time, corresponding to new capacity needs. Logical volumes are accessed as block devices named /dev/VolumeGroupName/LogicalVolumeName. For folks who have been working with commercial Unices, LVM support was a badly needed feature. Along with [this] (although not yet in the main kernel sources) comes the need for a filesystem type that can be expanded on the fly. Patches for IBM's JFS (Journaled File System) are available for both the 2.2 and 2.4 versions of the kernel, and ReiserFS, while currently focused on performance (i.e. burying ext2fs, the current status quo). Both of these play heavily into Linux's future, and hopefully will be integrated into the main kernel before 2.4.1 is released.