Presently, there are several leading companies on the market that offer hard drive storage devices. They can be divided - according to capacity, possibility of being scaled and performance - into two groups: storage systems devised for mid-performance computing platforms and those designed for high-performance. In capacity they range from around tens of gigabytes to several terabytes. All the most commonly used server platforms and operation systems can be connected to such hard drive devices.
Manufacturers are implementing into their products various hardware technologies and are taking differing approaches to solving other copying services. An alliance was created by IBM and Compaq this summer on this market. That alliance has produced the Modular Storage Server, designed mostly for PC servers and mid-performance Unix systems.
External hard drive storage devices make use of discs with a capacity of 9 to 50GB and a frequency of 10,000 rpm, with the use of SCSI UltraWide-3, SSA (Serial Storage Architecture) technologies, or alternatively fibre channel technology.
As far as preventing independent disc failures, manufacturers use RAID (Redundant Array of Independent Discs) technologies. Also often used is the hot spare or online spare discs technology, which immediately after detecting a failure of an active disc can take over its activities.
Interesting for any end-user might be a comparison of the size of the offered user disc space and the actual disc space that he/she has to purchase. This parameter may differ from manufacturer to manufacturer, and is further influenced by the enlarging copying services that are provided by these storage devices. They can include flash copy (snap shot), or the possibilty of synchronised or non-synchronised mirroring of the storage area by another storage area of the same type.
The performance of these devices depends on the level of performance, intelligence and optimalisation of the firmware, on the size of the read and write cache and optimalisation of the algorithm in the cache memory. Another influential factor is load balancing within the storage device, or communication with servers connected through external interface.
One of the important aspects that needs to be taken into consideration when evaluating such storage devices is the redundance of separate elements related to the resistance of the storage device to powercuts. Naturally, it would be useful for the storage device to have cache memory mirrored by a powerful battery, use self-correcting algorythms for repeated errors and to have power sources protected against short-term powercuts. Some contemporary storage devices are able to retain data in cache memory for up to seven days.
As far as provisions for a quick reaction to an error are concerned, some devices are equipped with HW and SW call home support. This means that when an error occurs, it is not only recorded into the error log, but is reported through a modem to the service centre of the provider. A service technician will then analyse the error and, after consulting with the client, eliminate it.
Stanislav Dzurík is an IT consultant at Columbex International. Comments and questions can be sent to his email address at: firstname.lastname@example.org
16. Oct 2000 at 0:00 | Stanislav Dzurík