After you’ve successfully installed a hard drive,
you must perform two more steps to translate a
drive’s geometry and circuits into something usable to the system: partitioning and formatting. Partitioning is the process of electronically subdividing the physical hard drive into groups of cylinders called partitions (or volumes ) A hard drive must have at least one partition, and you can create multiple partitions on a single hard drive if you wish. In Windows, each of these partitions typically is assigned a drive letter such as C: or D:. After partitioning, you must format the drive. Formatting installs a file system onto the drive that organizes each partition in such a way that the operating system can store files and folders on the drive. Several different types of file systems are used in the Windows world. Partitions provide tremendous flexibility in hard drive organization. Partitions enable you to organize a drive in a way that suits your personal taste. Partitioning enables a single hard drive to store more than one operating system (OS). One OS could be stored in one partition and another OS stored in a second. Windows 2000 and XP support two different partitioning methods: the older but more universal master boot record (MBR) partitioning scheme and the newer (but proprietary to Microsoft) dynamic storage partitioning scheme. Microsoft calls a hard drive that uses the MBR partitioning scheme a basic disk and a drive using the dynamic storage partitioning scheme a dynamic disk . Basic disk partitioning creates two very small data structures on a drive, the master boot record (MBR) and a partition table , and stores them on the first sector of the hard drive—called the boot sector . The MBR is nothing more than a tiny bit of code that takes control of the boot process from the system BIOS. When the computer boots to a hard drive, BIOS automatically looks for MBR code on the boot sector. The MBR has only one job: to look in the partition table for a partition with a valid operating system Allbasic disk partition tables support up to four partitions. The partition table supports two types of partitions: primary partitions and extended partitions. Primary partitions are designed to support bootable operating systems. Extended partitions are not bootable. A single basic disk may have up to three primary partitions and one extended partition. If you do not have an extended partition, you may have up to four primary partitions. Each partition must have some unique identifier to enable users to recognize it as an individual partition. Microsoft operating systems (DOS and Windows) traditionally assign primary partitions a drive letter from C: to Z:. Extended partitions do not get drive letters. After you create an extended partition, you must create logical drives within that extended partition. A logical drive traditionally gets a drive letter from D: to Z:. (The drive letter C: is always reserved for the first primary partition in a Windows PC.) Every primary partition on a single drive has a special setting called active stored in the partition table. This setting is either on or off on each primary partition. At boot, the MBR uses the active setting in the partition table to determine which primary partition to choose to try to load an OS. Only one partition at a time can be the active partition , because you can run only one OS at a time. The boot sector at the beginning of the hard drive isn’t the only special sector on a hard drive. The first sector of the first cylinder of each partition also has a special sector called the volume boot sector Primary Partition • If you want to boot an operating system from a hard drive, that hard drive must have a primary partition. The MBR checks the partition table for the active primary partition. In Windows 2000/XP, the primary partition is C:, and that cannot be changed. Active Partition • When you create a primary partition and decide to place an OS on that partition, you must set that partition as active. This must take place even if you use only a single primary partition. Luckily, this step is automated in the Windows installation process. Extended Partition • The first versions of the old DOS operating system to support hard drives only supported primary partitions up to 32 MB. • As hard drives went past 32 MB, Microsoft needed a way to support them. Instead of rewriting DOS to handle larger drives, Microsoft developers created the idea of the extended partition. That way, if you had a hard drive larger than 32 MB, you could make a 32-MB primary partition and the rest of the drive an extended partition. Extended Partition • The beauty of an extended partition is in the way it handles drive letters. When you create a primary partition, it gets a drive letter and that’s it. But when you create an extended partition, it does not automatically get a drive letter. Instead, you then go through a second step, where you divide the extended partition into one or more logical drives. An extended partition may have as many logical drives as you wish. By default, Windows gives each logical drive in an extended partition a drive letter and most Windows users use drive letters. With the introduction of Windows 2000, Microsoft defined an entirely new type of partitioning called dynamic storage partitioning, better known as dynamic disks. Dynamic disks drop the word partition and instead use the term volume. There is no dynamic disk equivalent to primary vs. extended partitions. A volume is still technically a partition, but it can do things a regular partition cannot do, such as spanning. A spanned volume goes across more than one drive. Windows allows you to span up to 32 drives under a single volume. Dynamic disks also support RAID 0 in Windows 2000 Professional and Windows XP Professional. Windows 2000 Server and Windows Server 2003 support RAID 0, 1, and 5. Dynamic disks use an MBR and a partition table, but these older structures are there only for backward compatibility. All of the information about a dynamic disk is stored in a hidden partition that takes up the last 1 MBof the hard drive Youcan use five volume types with dynamic disks: simple, spanned, striped, mirrored, and RAID 5. Most folks stick with simple volumes. • Simple volumes work much like primary partitions. If you have a hard drive and you want to make half of it C: and the other half D:, you create two volumes on a dynamic disk. • Spanned volumes use unallocated space on multiple drives to create a single volume. Spanned volumes are a bit risky⎯if any of the spanned drives fails, the entire volume is permanently lost. • Striped volumes are RAID 0 volumes. You may take any two unallocated spaces on two separate hard drives and stripe them. But again, if either drive fails, you lose all your data. • Mirrored volumes are RAID 1 volumes. You may take any two unallocated spaces on two separate hard drives and mirror them. If one of the two mirrored drives fails, the other will keep running. • RAID 5 volumes , as the name implies, are for RAID 5 arrays. A RAID 5 volume requires three or more dynamic disks with equal-sized unallocated spaces. The partition types supported by Windows are not the only partition types you may encounter— other types exist. One of the most common is called the hidden partition. A hidden partition is really just a primary partition that is hidden from your operating system. Only special BIOS tools may access a hidden partition. Hidden partitions are used by some PC makers to hide a backup copy of an installed OS that you can use to restore your system if you accidentally trash it—by, for example, learning about partitions and using a partitioning program incorrectly. A swap partition is another special type of partition, but it is only found on Linux and BSD systems. A swap partition is an entire partition whose only job is to act like RAM when your system needs more RAM than you have installed. Windows has a similar function called a page file that uses a special file instead of a partition. Most OS experts believe a swap partition is a little bit faster than a page file. Partitioning is not a common task. The two most common situations likely to require partitioning are when you’re installing an OS on a new system, and when you are adding a second drive to an existing system. Each version of Windows offers a different tool for partitioning hard drives. For more than 20 years, through the days of DOS and early Windows (up to Windows Me), you used a command-line program called FDISK to partition drives. Windows 2000 and Windows XP use a graphical partitioning program called Disk Management . Linux uses a number of different tools for partitioning. The oldest is called FDISK⎯yup, the exact same name as the DOS/Windows version. However, that’s where the similarities end, as Linux FDISK has a totally different command set. Even though every copy of Linux comes with the Linux FDISK, it’s rarely used because so many better partitioning tools are available. One of the newer Linux partitioning tools is called GParted. GParted is graphical like Disk Management and is fairly easy to use (Figure 9.8). GParted is also a powerful partition management tool⎯so powerful that it also works with Windows partitions. Before Windows 2000, there was no way to do this nondestructively. As a result, a few third-party tools, led by Symantec’s now famous PartitionMagic, gave techs the tools to resize partitions without losing the data they held. Windows 2000 and XP can non-destructively resize a partition to be larger but not smaller. Vista lets users non-destructively resize partitions any way they wish! RAID stands for Random Array of Independent Device and is the process of making two or more drives look like one drive. Three appreciated versions: • RAID 0 Data is split between two drives. Fast, but if one drive fails, all data is lost. Also called striping. • RAID 1 All data is copied to two drives, with each drive holding an exact copy of the other. Slow, but if one drive fails, the other drive has all the data. Also called mirroring. • RAID 5 Data and special recovery information are spread across three or more drives. If one drive fails, the other two drives can recover all data. Also called striping with parity.