Hard Disks and Bottlenecking



Hard disks used in servers are typically different than those used in desktops. The speed that the platters spin can significantly affect the performance of the entire server. Some of the various performance parameters used in hard drives are transfer rates, seek times, and latency. A hard disks latency is measured in milliseconds and includes both rotational latency and seek time; which is the total time it takes a hard drive’s read/write head to locate the location on the disk. A driver’s access time is also worth mentioning; which is the interval between the time a request for data is made by the system and the measurement of time that data is then made accessible from the drive. A hard drive’s transfer rates are the speed that data is sent to and from the disk platter and is another crucial parameter. These rates are usually measured in MBps. Generally, for a server, higher-performance hard disks are only necessary if the server will be running advanced applications, such as databases. Some entry-level servers offer hot-swappable drives which enable a hard drive to be replaced without shutting anything down or opening the case.
            A hard disk that is bottlenecking can significantly impact the entire server. While the first fix that comes to mind is just installing a different disk, the optimal solution depends on the actual cause of the bottleneck. Ruling out memory storage problems should be the first step. This can be fixed by installing a larger disk, compressing the original disk, or by creating disk quotas to restrict the growth of user files. Defragmenting the disk is another useful thing to do and can be done using the Disk Defragmenter tool. Some other tips are to place multiple drives on separate I/O buses and to distribute the workload among multiple drives.
            Regarding the proper disk specifications, one should strive to obtain for their server, using the highest level of intelligent and efficient components available to you is always recommended. These include the controller, I/O bus, the actual disk, and the cabling used. Upgrading to higher-speed and wider-bandwidth components are also recommended which will typically decrease transfer time and increase throughput. Some other areas to monitor in regards to bottlenecks are queue depth, average I/O size, IOPS, and write vs. read percentages.
            While it may seem that doubling CPU speed may significantly increase the performance of the overall storage, this is not always the case. This is due to the many other factors inside and outside of the CPU that significantly affects the overall system performance. As an example, a 2GHz CPU will not necessarily be twice as fast as 1 GHz CPU because it depends on the quantity of work each CPU completes in each clock cycle. The 1GHz CPU might actually be faster if it is more efficient and can complete more tasks in each cycle, depending on its architecture. The number of cores a CPU has is what greatly increases the performance, as well as its cache size, the speed of the front side bus, amount of system memory, as well as overclocking capabilities.
Sliwa, Carol.Tech Target. Searchstorage.techtarget.com. Web. Retrieved 26 Jun 2017.

Tech-FAQ. Tech-faq.com. Web. Retrieved 26 June 2017.




Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.