Optimising Disk Performance
Although hard disk speeds have improved immensely, they are still mechanical devices, and as such are some of the slowest components to be found in a computer system. Servers make intensive use of disk read and write operations, so even a fairly minor improvement in speed can offer significant performance benefits. There are three major ways of improving the read/write performance of disks on a Windows Server 2003 system:
- Implement disk striping (RAID-0)
- Implement disk striping with parity (RAID-5)
- Defragment the disks regularly
RAID stands for Redundant (or Random) Array of Inexpensive Disks. Various RAID configurations exist, but only RAID-0, RAID-1 and RAID-5 are now in common use.
In RAID-0 (disk striping) data is written to 2 or more disks simultaneously, eg: the start of a file may be written to one disk, the middle of the file is to a second disk and the end of the file to a third disk, at the same time. Conversely, the start, middle and end of a file can be read simultaneously from all the disks. Carrying out these operations in parallel produces major performance improvements. However, if one of comprise the striped (RAID-0) volume fails, all data on that volume is lost, so RAID-0 should only be used on volumes that need to be read and written quickly but that do not require fault tolerance.
RAID-5 (disk striping with parity) is the safest way to increase the read/write performance of disks a it also provides fault tolerance. If one of the disks that comprise the RAID-5 volume is lost, the data can be recovered by using the parity information stored on the other disks. In Windows Server 2003, RAID-5 can easily be implemented via the Disk Management console, using three or more dynamic disks.
Fragmentation occurs naturally on hard disk drives as files are written and deleted. Hard disks allocate free space as it becomes available, so a newly written file may be stored as several different chunks at various locations across the hard disk drive. This means that it takes longer to read or write the file, as the read-write heads have to move to several different locations. When a file is defragmented, all the chunks are rearranged on the disk so that they are contiguous. This can be done using the Disk Defragmenter console or the defrag command-line utility.
|The activities given on the following pages can contribute towards your log book for this unit. You must demonstrate that you can optimise server disk performance by implementing RAID solutions and defragmenting volumes and partitions.|
Next: RAID-0 Volumes