RAID 10 was not mentioned in the original 1988 article that defined RAID 1 through RAID 5. The term is now used to mean the combination of RAID 0 (striping) and RAID 1 (mirroring). Disks are mirrored in pairs for redundancy and improved performance, then data is striped across multiple disks for maximum performance. In the diagram below, Disks 0 & 2 and Disks 1 & 3 are mirrored pairs.
Obviously, RAID 10 uses more disk space to provide redundant data than RAID 5. However, it also provides a performance advantage by reading from all disks in parallel while eliminating the write penalty of RAID 5. In addition, RAID 10 gives better performance than RAID 5 while a failed drive remains unreplaced.
Under RAID 5, each attempted read of the failed drive can be performed only by reading all of the other disks. On RAID 10, a failed disk can be recovered by a single read of its mirrored pair.
Figure 14
7.3 Summary
As processor and main memory speeds increase more rapidly than those of secondary storage devices, optimizing disk performance has become important to realize optimal performance. As the platters of disk spin, each read-write head sketches out a circular track of data on a disk surface to access. The time it takes for the head to move from its current cylinder to the one containing the data record being accessed is called the seek time. The time it takes for data to rotate
from its current position to a position adjacent to the read/write head is called latency time. Many processes can generate requests for reading and writing data on a disk simultaneously. Because these processes sometimes make requests faster than they can be serviced by the disk, queues build up to hold disk requests. Some early computing systems simply serviced these requests on a first-come-first-served (FCFS) basis which is a fair method, but when the request rate becomes heavy, FCFS results in long waiting times. To reduce the time spent seeking records, it seems reasonable to order the request queue in some other manner. This reordering is called disk scheduling. The two most common types of scheduling are seek optimization and rotational optimization which are evaluated by comparing their throughput, mean response time and variance of response times.
Shortest-seek-time-first (SSTF) scheduling services the request that is closest to the read-write head's current cylinder. By reducing average seek times, SSTF achieves higher throughput rates than FCFS, and mean response times tend to be lower for moderate loads. The SCAN scheduling strategy reduces unfairness and variance of response times by choosing the request that requires the shortest seek distance in a preferred direction. Thus, if the preferred direction is currently outward, the SCAN strategy chooses the shortest seek distance in the outward direction. However, because SCAN ensures that all requests in a given direction will be serviced before the requests in the opposite direction, it offers a lower variance of response times than SSTF.
The circular SCAN (C-SCAN) modification to the SCAN strategy moves the arm from the outer cylinder to the inner cylinder, servicing requests on a shortest-seek basis. When the arm has completed its inward sweep, it jumps to the outermost cylinder, then resumes its inward sweep processing requests. C-SCAN maintains high levels of throughput while further limiting variance of response times by avoiding the discrimination against the innermost and outermost cylinders.
N
The FSCAN and -Step SCAN modifications to the SCAN strategy eliminate the possibility of indefinitely postponing requests. FSCAN uses the SCAN strategy to
service only those requests waiting when a particular sweep begins. Requests arriving during a sweep are grouped together and ordered for optimum service during the return sweep. N-Step SCAN services the first n requests in the queue using the SCAN strategy. When the sweep is complete, the next n requests are serviced. Arriving requests are placed at the end of the request queue, which prevents requests in the current sweep from being indefinitely postponed. The LOOK variation of the SCAN strategy "looks" ahead to the end of the current sweep to determine the next request to service. If there are no more requests in the current direction, LOOK changes the preferred direction and begins the next sweep, stopping when passing a cylinder that corresponds to a request in the queue. This strategy eliminates unnecessary seek operations experienced by other variations of the SCAN strategy by preventing the read/write head from moving to the innermost or outermost cylinders unless it is servicing a request to those locations.
The circular LOOK (C-LOOK) variation of the LOOK strategy uses the same technique as C-SCAN to reduce the bias against requests located at extreme ends of the platters. When there are no more requests on a current sweep, the read/write head moves to the request closest to the outer cylinder and begins the next sweep. Sector queuing is a scheduling algorithm for fixed head devices such as drums.
RAID is the organization of multiple disks to achieve higher data transfer rates on large data accesses and higher I/O rates on small data accesses. Data striping also results in uniform load balancing across all of the disks. To cope with the problem of reliability, redundancy is introduced which has its own limitations at the time of write operations. A number of RAID organizations are available with their merits and demerits.
7.4 Keywords
¾ Seek time: To access a block from the disk, first of all the system has to move the read/write head to the required position. The time consumed in moving the head to access a block from the disk is known as seek time.
¾ Latency time: the time consumed in rotating the disk to move the desired block under the read/write head.
¾ Transfer time: The time consumed in transferring the data from the disk to the main memory is known as transfer time.
¾ RAID: It is the Redundant Array of Inexpensive Disks, an organization of multiple disks into a large, high performance logical disk.