|
While there is little dispute among IT professionals regarding the impact of disk fragmentation on system performance, no independent guidelines exist to recommend the frequency of defragmentation across an infrastructure. Some IT professionals use defragmentation as a measure of last resort, defragmenting only after system performance has sufficiently degraded to make its impact directly noticeable to users. Others proactively schedule disk defragmentation regularly, with the intent of eliminating the gradual accumulation of fragmented files.
While just about every IT professional has his or her fragmentation horror story ? about a system on which fragmentation had so severely degraded performance that it was unusable ? very few can offer more than anecdotal evidence regarding the use and importance of defragmentation software.
I recently decided to put fragmentation to the test, with the intent of answering two basic questions:
1. What impact does fragmentation have on user and system activities?
2. How quickly does fragmentation accumulate as a result of these activities?
This white paper will outline the results of the testing, draw conclusions, and make recommendations regarding managing fragmentation across your infrastructure.
The Mechanics of Disks
The basic components of hard disks (see Figure 1) have not changed significantly since their invention in the 1950s. Hard disks have one or more polished platters made of aluminum or glass that hold a magnetic medium used for storing information. The platters are stacked onto a spindle and rotated by a spindle motor at very high speeds, often in excess of 160 miles per hour. A platter has concentric circles called tracks, and each track is divided into small sections called sectors, each capable of holding a fixed amount of information.
Small devices called heads are responsible for the actual reading and writing of data on the platter. Each platter has two heads (for the top and bottom), and the heads are mounted on sliders positioned over the surface of the disks, which in turn are mounted on arms. The entire assembly is connected to and controlled by an actuator, which in turn is connected to a logic board that allows for the communication between a computer and the hard disk.
To read or write information to the disk, an application makes a request of an operating system to create, modify or delete a file. The operating system then translates the logical request into a physical request containing the actual locations to be read or written on the hard disk. The logic board then instructs the actuator to move the heads to the appropriate track, and to read or write the appropriate sectors from the rotating platter below.
The mechanical movement of the head across a platter is typically one of the most expensive operations of a hard disk. As a result, most operating systems seek to minimize this head movement through caching, optimizing I/O requests, and streamlining the storage of data on a disk. Streamlining the storage of data typically involves writing the data for individual files in a file system contiguously on a platter, allowing the head to read or write data without needing to be repositioned.
Due to their mechanical nature, hard disks represent one of the poorest-performing components in a system. Electronic components, such as the CPU, motherboard, and memory, are improving performance at a much faster pace than hard disks, whose performance is limited by the mechanics of spinning a platter and moving a head. As a result, since an integrated system is often as fast as its slowest component, it is essential to ensure hard disks are performing at their optimum level.
NTFS
Before discussing fragmentation, let's discuss briefly the New Technology File System (NTFS), the file system used with all modern versions of Windows and the focus of the testing for this white paper. While understanding a specific file system is not a pre-requisite to understanding fragmentation, it will help clarify both the terminology used as well as the test results.
NTFS was created by Microsoft in the 1990s as part of its strategy to deliver a high-quality, high-performance operating system capable of competing with UNIX in a corporate environment. NTFS divides a hard disk into a series of logical clusters whose size is determined at the time the disk is formatted with the file system.
|
