Articles and Information
by Members for Members

Hand-in-Hand: File Systems and Hard Drives

by Lee Alexander

General

As with most components of a computer system there are two aspects: hardware and software. In order to have a "permanent" memory, your PC employs a Hard Drive and a File System. We put “permanent" in quotes because things can happen, even tombstones can wear out. Where do we get the term Hard Drive? The early non-volatile memory media were referred to as Floppy Disks, soon to be (if not already) obsolete. The later versions of floppy disks were 3.5 inch flexible plastic discs encased in a hard plastic case. By nonvolatile memory we mean the data remained when the power was removed.

A hard drive resembles a mini juke box, consisting of multiple platters and read/write heads analogous to vinyl records and phonograph pickups. The "hard" platter is typically a highly polished aluminum disk coated with a very thin magnetic film on both sides. Instead of a single pickup, there are two read/write heads for each disk, one for each side. Disks are stacked on a single spindle and all rotate at the same speed. Rotational speeds vary from 4500 to 15,000 rpm; today's common standard is 7200 rpm.

Formatting

As a verb, to format has multiple interpretations: in graphics we can change the format of an image (BMP, JPG, TIFF, etc.), we can change the look of a document's format (font, margins, paragraph layout, etc.). Our topic is Hard Drives and that restricts our formatting to the physical and electronic properties of the disks.

There are two levels of formatting: Low Level and High Level. Low Level formatting defines the tracks and sectors (and thus the cylinders) of the stacked platters. This is done by the manufacturer (I will forego the formatting of floppies). The electronics, at this point, will label bad sectors (nothing is perfect) to avoid writing data to these areas.

High Level formatting defines the areas on the disk as to cylinders and sectors and creates a file allocation table, FAT, to assign addresses to clusters. A cluster is the smallest unit of the HD that can be addressed. It is composed of sectors, how many depends upon the File System. More on that subject, later. Cylinders consist of tracks stacked on the disks as they rotate in unison. To see these parameters go to Start | Run and enter msinfo32.exe . In the left hand pane, expand Components, Storage, and click on Disks.

Communication Software

The connection between the HD and the computer is over a bus – ATA (Advanced Technology Attachment). As the technical aspects of HDs have gone through generational changes, so has the ATA Bus. ATA also has a few aliases: IDE (Integrated Drive Electronics) has control features built into the drive, EIDE (Enhanced IDE), DMA (Direct Memory Access), or Ultra DMA. The first version, ATA-1, was a 16-bit parallel connection introduced by Compaq in 1986. It supported two devices on the same connection and had a transfer (read/write) speed of 4 MBps (Mega Bytes per second). Jumping ahead, ATA-4 introduced ATAPI (ATA Protocol Interface) which allowed other devices (CD and DVD drives) to coexist on the same bus. ATA-5 gave us a rate of 66.6 MBps.

However, here is a note of warning. For ATA bus connections, if two devices are on the same bus, the speed will be dictated by the slower device. Most PCs have two buses and drive controllers. Use this feature to separate fast hard drives from slower removable media drives.

On the way to ATA-6, expected to be capable of addressing more than 137 GB at a speed greater than 100 MBps, we get SATA – Serial ATA. Counter-intuitively, we get higher transfer rates with a single conductor than with a parallel multi-conductor ribbon cable.

RAID

The acronym originally stood for Redundant Array of Inexpensive Disks; then it became Redundant Array of Independent Disks; and how, Redundant Array of Independent Devices. Once the province of enterprise systems, this feature has now propagated to home PCs, admittedly highline units. The basic concept of RAID was twofold: to surmount the expense of large capacity hard drives and address limitations, and to provide data protection by mirroring data on separate physical drives. Raid comes in a variety of configurations:

•  RAID 0 – “striping” - two or more physical drives are written sequentially. The array of drives appears as a single large logical disk to the computer.

•  RAID 1 – “mirroring” - two physical drives receive the same data, essentially backing up each other.

•  RAID 3 - involves error correcting technology, requiring a minimum of three drives. Two drives are used as in RAID 0, interlacing data to create a larger volume. The third drive stores CRC (Cyclic Redundancy Check) checksums. A checksum is a number calculated by an algorithm that validates a block of data. The beep you hear in a supermarket, as an item is scanned and using a checksum algorithm, is confirmation that the bar code was read successfully.

•  RAID 5- also requires a minimum of three physical drives (although most systems use five drives). Generally used with large, low-fault-tolerant databases, it differs from RAID 3 in that all data and error correcting codes are distributed among all drives.

For home computers, the most commonly used system is RAID 0 or RAID 1.

There are two methods for implementing RAID - hardware and software. The hardware implementation uses a special disk controller with its own CPU to control traffic. The array of hard drives appears to the computer as a single logical drive. The drives can be external to the host computer (in their own case) and connected via a network connection or special interface card.

In the case of software RAID, the array is created through special software. As an application, it occupies memory on the host system, consumes CPU cycles, and is system-dependent in operation. The host computer will take a performance hit when using software RAID.

File Systems

Windows presently has three File Systems: FAT (actually FAT 16) [File Allocation Table] using a 16 bit word address scheme; FAT32 using 32 bits; and NTFS [New Technology File System] a significant improvement over the other schemes. In the binary machine language of computers, a 16 bit word – 2 16  – affords 65,536 distinct numbers for addresses. A 32 bit word – 2 32 – allows for 4,294,967,296 [4 + billion] addresses.

NTFS is a very complicated system that removes most of the restrictions of FAT systems. As a Novice topic, we will try to cover just the principal advantages of NTFS over FAT. If you crave more details try www.digit-life.com/articles/ntfs for an article by Dmitry Mikhailov. The English is a little fractured but the facts and sense are all there.

FAT is available for MS-DOS and all versions of Windows, FAT32 is available for OS's starting with Windows 95 OSR2. NTFS is only available for Win XP and Win 2000. For Win NT SP4 and later, it is a hit or miss proposition. The only reason not to us NTFS is for backward compatibility with older files. The earlier file systems never anticipated today's HD capacities nor the use to which we would put our PCs storage. FAT32's limit of a volume size of 2 TB (Terabytes, 1TB=1,000 GB) was deemed almost limitless. Digital audio, imaging, photography, and video can consume Gigabytes of storage like army ants at a picnic.

Clusters