Choosing a File System: FAT, FAT32, or NTFS?
The next major consideration on the pre-installation agenda is determining what type of file system you intend to use. The rules and regulations discussed in the preceding section might have narrowed down this choice for you. Still, you'll likely want to read about the pros and cons of the various file-keeping schemes in use on Windows XP machines and consider a few details on how they influence your installation.
Windows XP doesn't know about compressed drives such as those created with DriveSpace or DoubleSpace. You have to decompress them before installing to Windows XP. Decompressing is a real pain if your disk space is totally packed (because as your files decompress, they fill up even more space on an already cramped drive). You might have to decompress in stages, moving data off the hard disk to backup media or another drive.
As mentioned in the preceding sections on dual-booting Windows XP and a second operating system, it's a good idea to think about what file system you're going to use, preferably before installing Windows XP. Although you can use utilities in Windows XP and external utilities such as PartitionMagic to convert partitions between file systems after the fact, forethought and advance partition preparation are the better path. Let's do a little review of file systems you can use and advantages of each.
A file system is a scheme by which data files and directories (folders) are stored and retrieved on a floppy disk or hard disk. Tape and other media have file systems as well, but here I'm talking only about hard disks. Windows XP supports three file systems: the NT File System (NTFS) or one of the file allocation table file systems (FAT or FAT32).
A Brief History of File Systems
In the beginning, there was FAT, and it was good. FAT is the system that DOS uses; it's been around for a long time, since the early '80s. FAT stands for File Allocation Table. A file allocation table is basically a table of contents of the disk that the operating system uses to look up the location of a file, even if the file is broken up in pieces (sectors) scattered across the disk's surface. The FAT scheme brought relatively simple, reliable, and efficient floppy and small hard disk storage to the PC. It's also the scheme that, unlike the Macintosh file system, brought the confining 8.3 file-naming convention that many of us learned to live with and hate. For example, myletter.doc is an example of the longest-possible 8.3 filename.
When NT 3.x appeared, it included NTFS as an acknowledgment of the shortcomings of the FAT system, including a hard drive partition size limit of 2 gigabytes (GB). NTFS provided long filenames, more security and fault tolerance, better disk compression, support for hard disks up to 2 terabytes (that's big), and support for advanced multiple-disk arrangements such as striping and mirroring (RAID). Also, as drives become larger, efficiency of disk storage doesn't fall off under NTFS as it does with FAT.
Windows 95 brought long filenames to FAT through some sleight of hand, but still the system was not good at dealing with the newer large drives and wasted a bunch of space on them when it stored tiny files. So, to both provide good backward compatibility with FAT disks and still offer support for large drives, Windows 95 OSR2 and Windows 98 both included a new file system called FAT32. Essentially a beefed up FAT file system, FAT32 isn't as robust as NTFS, and it's not compatible with NTFS. The FAT32 system eliminated the 2GB upper limit on partition size support (it also can run as high as 2TB) and increased effective storage capacity by lowering the cluster size on large drives. NT 4 can't read or work with FAT32, nor can DOS and Windows 3.x. However, both Windows 2000 and Windows XP support FAT32.
Clusters are the smallest amount of space that can be used to store information on a hard disk. On smallish drives, the cluster size is also pretty small, so storing dinky files that are only, say, 1KB in size is pretty efficient (there is little wasted space). But on today's huge drives, under the FAT scheme, the cluster would necessarily be much larger (this is a limitation of the FAT system, not the drive). You end up donating serious amounts of space to no good cause. The bottom line is that FAT32 and NTFS get you more bang for your hard disk buck, because lowering the cluster size allows for more efficient use of space, especially on larger drives.
The bottom line? As with Windows NT and Windows 2000, NTFS is the recommended file system for use with Windows XP. NTFS v5 has all the basic capabilities of FAT as well as all the advantages of FAT32 file systems. The weird thing is that now you have to think about three different file systems when considering dual-booting. When you consider that you have at least nine Microsoft operating systems to choose from and three file systems, the combinations get complex. Therefore, understanding the limitations of each is important.
You can convert an existing partition to NTFS during setup, but if you want to wait, you can convert it later by using a command-line utility called convert.exe (see "Convert" in Chapter 29). Another approach is to use PartitionMagic, which is discussed later in this chapter.
One of the prime points to remember is that if you're dual-booting, only Windows NT, Windows 2000, and Windows XP systems can read NTFS partitions. If you don't care about accessing the NTFS partition from, say, Windows 98, this is not a big deal. It simply does not appear in Windows Explorer and is not available from your applications. This is the main reason Microsoft doesn't want you to mix file systems; it simply confuses people.
When you're running partitions larger than 32GB, you should really format them as NTFS. If you choose to use FAT, anything over 2GB should be formatted FAT32.
File systems are terribly complex and a subject far beyond what I can cover in a Windows book. For the most part, you don't need to know more than what is presented here, unless you are naturally inclined to learn everything you can about complex topics (don't worry, you are not alone). In that case, I recommend that you pick up a copy of Upgrading and Repairing PCs, 13th Edition, by Scott Mueller (also published by Que).
In case you don't know, disk partitioning is a scheme by which you can have a single hard disk look like multiple hard disks to the operating system. If you partition a disk into, say, two partitions, the operating system displays disks C and D rather than just C. You split up the space on the drive between the partitions based on your needs.
One of the most notable needs for disk partitioning was to accommodate operating systems that imposed limitations on the size of partitions. As hard disks grew in size, partitioning was required in order to use the entire disk. Because FAT had a limitation of 2GB, users relied on partitioning or other software driver schemes to get around this imposed top end. Another common reason for partitioning is for running dissimilar operating systems, ones that cannot read from or write to a common file system. Because each partition can have its own disk format, this could often circumambulate such requirements. One partition could be FAT, another NTFS, and another HPFS (for OS/2), and so on. Any hard disk can contain up to four primary partitions.
Only primary partitions can be marked active. An active primary partition is where the computer's BIOS looks for a bootable operating system when powered on. In addition to primary partitions, there are extended partitions. A system can have up to four primary partitions or a maximum of three primary partitions and a single extended partition. The extended partition is a bit different than a primary partition. First, extended partitions cannot be marked active. Second, extended partitions must be divided into logical drives. Only primary partitions and logical drives can be formatted and assigned a drive letter. To total number of primary partitions plus logical drives cannot exceed 32 for a single hard drive.
For examples of these issues, see Figure 3.1. Disk 0 has a single primary partition (C), and an extended partition. The extended partition contains two logical drives (D and G) and 1.86GB of free space. The extended partition is highlighted with a thick border; that's how you distinguish it from primary partitions. This extended partition could contain another 29 logical drives before reaching the 32 division maximum per drive. Disk 1 consists of a single primary partition (F) and 15.11GB of free space. This free space could be used to create up to 3 additional primary partitions, or up to 2 additional primary partitions and an extended partition containing logical drives. Drive 2 consists of a single primary partition (E).
Figure 3.1 Disk Management viewed through Computer Management, showing drive partitioning.
In most cases, you'll rarely need to divide a drive into more than 4 sections. And then, that may only occur when you need to divide a large drive into FAT partitions (because each partition cannot be larger than 2GB). The tools you use to create partitions will manage the division classification for you. If you use FDISK, you'll be a bit more aware of the presence of extended partitions and logical drives. If you use the Windows NT, Windows 2000, or Windows XP drive tools (such as the Disk Management tool) you'll see extended partitions and logical drives labeled, but the OS manages when these need to be created.
Some people use partitions for dividing up their data rather than for accommodating different file systems. You might want a partition to organize informationfor example, one for backup data, one for documents and data files, one for applications only, or for the operating system only. Then you can more easily design your backup strategy.
When you do a new installation of Windows XP, the Setup program looks around and automatically selects an appropriate disk partition as the destination based on size and format. You can override the choice by clicking the Advanced Options button during Setup, though.
If you're going to dual-boot, you should install on a separate partition. Either create one or use one that is already present. The reason for using a separate partition for each operating system is to prevent Setup from overwriting important files belonging to the other operating system. If you have unpartitioned (different from unused) space on your disk, Setup can create a partition during installation.
If you intend to dual-boot Windows 9x and Windows XP, Windows 9x should be on the first partition. In general computer terms, this means the boot drive. In Windows NT speak, this is the system partition.
→ If you're considering creating a system that is bootable in more than one operating system, see "Dual-Booting Versus Single Booting," p. 75.
The exact options you have during Setup change depending on your existing hard disk configuration. You might have as many as four options when partitioning your hard disk:
If you have adequate unpartitioned space, you can create a new Windows XP partition in that space and install Windows XP in it.
If the hard disk is unpartitioned (no partitions at allfreshly formatted) you can create and size the Windows XP partition.
If the disk does have an existing partition, but you don't care what's in it, Setup lets you delete the partition and create a new one of your chosen format for installing Windows XP. Beware, though; deleting an existing partition destroys every file on the partition.
If the existing partition is large enough, it's actually possible to put Windows XP on it. Contrary to what the Microsoft documentation says, installing on an existing partition doesn't overwrite all data on that partition. If you choose a directory separate from the existing operating system (for example, install into C:\winxp), Setup leaves any existing C:\windows directory alone. But don't forget, the \Program Files and \Documents and Settings folders may be altered by the new OS so that the pre-existing OS can't access the data stored there. Things can get kind of funky with other possible file collisions, too. You can dual-boot, but you might have to reinstall your applications. The better approach is to play it safe and create a new partition to install into.
Don't even try to install Windows XP on a partition that is less than 1GB. Although Windows XP only requires 650MB (only!), I think creating a 2GB partition is more reasonable. If you are upgrading you'll need an extra 300MB to store the recovery files, and Setup won't allow you to start the process with less than 1GB of free space. You'll want (and need) the additional free space later for future additions and modifications to the operating system.
File System Conversion Limitations
This section lists a few ridiculous warnings and limitationsridiculous because these steps are not required if you use PartitionMagic. Due to limitations of Setup and Microsoft's supplied disk tools (FDISK, for example), you should be aware of these issues. First, most conversions between file systems are multi-step processes involving backing up and then restoring the partitions after reformatting.
For example, although converting to NTFS during installation is easy, if you change your mind and want to revert to FAT, you have to back up all your files on the NTFS partition, reformat the partition as FAT (which erases all the files), and then restore the files from backup. The same is true of converting a FAT partition to FAT32. One workaround is to use the FAT32 converter in Windows 98 or 95 OSR2. Each of these operating systems has a tool that performs such a conversion quite easily.
Second, you can't restore an NTFS v4 (file system from NT 4) partition after you convert it to NTFS v5 (file system from Windows XP, Windows 2000, and Windows NT 4.0 SP4+). There's no easy way out on that score. There is support for NTFS v5 file formats via tools like PartitionMagic to do this, though.