Computers:Hard Drive
From DpWiki
Contents |
Introduction
Another critical component in an image processing machine, the hard drives are the repositories that perminantly store all programs and data that will be used by the machine. In addition, hard drives are also used as temporary working areas for tasks that require more space than is available in main memory.
Interface
ATA
Short for 'Advanced Technology Attachment', ATA is used to connect storage devices like optical drives (eg DVD Recorders) and hard drives. Using a parallel bus, the fastest version of this interface currently provides a maximum theoretical bandwidth of 133MB/s and can make use of up to two devices (bandwidth will be shared between them). To help distinguish between ATA and SATA, it is commont to refer to this interface as PATA (short for Parallel ATA).
With the introduction of SATA, hard drives have almost all shifted over to the new standard. The vast majority of optical drives still use this interface, however, so it is still relatively commonplace in modern computer systems.
SATA
The successor to PATA, SATA switches to a serial configuration that allows much smaller cables and higher bandwidth. The first iteration of SATA provides a theoretical bandwidth of 150MB/s, so it allows for faster opperation than conventional PATA. Unlike PATA, SATA only allows a single device to be connected to each bus. Fortunately, the smaller connector means that most controllers provide more available ports to connect devices to.
SATA2
A newer iteration of the SATA standard, adding support for a 300MB/s data rate, native command queueing and port multipliers. Note that some SATA I devices implement some of these features, however may not be SATA II compliant. Please see the linked articles for more information on each of these technologies.
SCSI/SAS
A long-standing interface used in high-end workstation and server systems, SCSI (Small Computer System Interface) is an extremely powerful interface. Controllers and drives built for this interface were designed for heavy use and provide many enterprise-level features not present in conventional PATA or SATA systems (eg 640MB/s throughput, 15,000RPM hard drives, support for 15 devices per cable, long cable runs, etc.). Unfortunately, its footing in the high-end space also means that it carries a significant price premium so it is not for everyone.
The older parallel SCSI bus is now being accompanied by the newer SAS (Serial Attached SCSI) interface. SAS uses the same physical interface as SATA and contains provisions for SAS controllers to support SATA hard drives being attached to them. This allows SAS-based systems to mix both types of hard drives as necessary.
Spindle Speed
The rate at which the platters inside of the drive rotate when the drive is opperational. Conventional hard drives generally rotate at 5400-7200 RPM. High speed destop drives go as fast as 10,000RPM and high-end workstation/server drives go up to 15,000RPM. As it takes less time for the disc to make a revolution, drives with faster spindles will generally provide higher throughput and shorter access times. This allows them to potentially provide significant benefits to overall performance.
Unfortunately, the faster the drive rotates the more difficult it becomes to increase the density of data stored on the disc. This generally means that high-speed drives will often have significantly smaller capacities than their slower bretheren. Also, decreased density can have negative effects on sustained throughput so it can counteract some of the gains brought about by the higher speed.
As such, the primary benefits of a drive with a high spindle speed is the decrease in access time. This means that for scenarios where the machine will be accessing many smaller bits of data (pagefiles/scratch disks, thumbnails, databases, etc.), these drives will generally show significant improvements in performance.
Access Time
The ammount of time it takes for the drive to repossition its head over a new block of data. As the drive is unable to retreive any data while this proces is taking place, this can have significant impact on some forms of data transfers. In effect, the access time is the overhead associated with accessing a new block of data.
When working with large files, access time is generally insignificant as the time taken to actually retreive the data is much higher. When working with many small files, however, this can easilly take more time than actually transferring the data so it becomes a significant factor.
Cache
In order to help speed up the operation of the system, modern hard drives are equipped with a small ammount of solid-state memory (usually 2-16MB) that acts as a cache. Like the cache in the main processor, this memory provides a high-speed location to temporarilly store data being transferred in and out of the hard drive. This, in turn, allows the system to work around many of the delays caused by the mechanical limitations of the hard drive.
Reliability
Hard drives are complex mechanical devices and, as such, are generally the component that is most prone to failure. As photographers will often store large collections of valuable image files on these drives, reliability is an important concern.
While drives designed for workstation or server duty ((WD Raptor and most SCSI offerings) often do provide MTTF ratings, most mainstream drives do not. As such, it is often difficult to gauge the robustness of the models being considered. Naturally, things like the length of the warantee and the cost of the drive are important indicators, its always a good idea to do some research into the experiences that other users have had with a specific product line.
RAID Arrays
As mentioned above, RAID controllers allow multiple hard drives to be logically combined to protect data or improve performance. See the RAID article for further information on this topic.
