13.1 through 13.5, 13.10 and 13.11

13.10 Parallelizing Disk Access Using RAID Technologyture for indexing. In general, any data structure that can be adapted to the characteristicsof disk devices can be used as a primary file organization for record placementon disk.icularECTs,pe. In:tes ofrshipstxamc'sthet freld5 Owncords,valuelentedchicalrysical,orbyIisk tophysirently,ncy atMENTent, itJDENT:'ing of-'d file.s fieldllv thevpe ofdeter-.ta valrple,ifis offert'e ivillstruc- $13.10 Parallelizing Disk AccessUsing RAID TechnologyWith the exponential growth in the performance and capacity of semiconductordevices and memories, faster microprocessors with larger and larger primary rxemoriesare continually becoming available. To match this growth, it is natural toexpect that secondary storage technology must also take steps to keep up withprocessor technology in performance and reliability.A major advance in secondary storage technology is represented by the developmentof RAID, which originally stood for RedundantArrays of Inexpensive Disks.Lately, the l in RAID is said to stand for Independent. The RAID idea received a verypositive industry endorsement and has been developed into an elaborate set ofalternative RAID architectures (RAID levels 0 through 6). We highlight the mainfeatures of the technology belowThe main goal of RAID is to even out the widely different rates of performanceimprovement of disks againsthose in memory and rnicroprocessors.l2 While RAMcapacities have quadrupled every two to three years, disk access times are improvingat less than l0 percent per year, and disk transfer rates are improving at roughly 20percent per year. Drsk capacities are indeed improving at more than 50 percent peryear, but the speed and access time inrprovements are of a much smaller magnitude.Thble 13.3 shows trends in disk technology in terms of 1993 parameter values andrates of improvement, as well as where these parameters are in 2003.A second qualitative disparity exists between the ability of special microprocessorsthat cater to new applications involving video, audio, image, and spatial data processing(see Chapters 24 and 29 for details of these applications), with correspondinglack of fast access to large, shared data sets.The natural solution is a large array of sn-rall independent disks acting as a singlehigher-performance logical disk. A concept called data striping is used, which utilizesparallellsnr to improve disk performance. Data striping distributes data transparentlyover multiple disks to make them appear as a single large, fast disk. Figure13.12 shows a file distriburedor striped over four disks. Striping improves overallI/O performance by aliowing multiple I/Os to be serviced in parallel, thus providinghigh overall transfer rates. Data striping also accomplishes load balancing amongdisks. Moreover, by storing redundar-rt information on disks using parity or someother error correction code, reliability can be improved. In Sections 13.3.1 and13.3.2, we discuss how RAID achieves the two important objectives of improvedreliability and higher performance. Section 13.3.3 discusses RAID organizations.'12. This was predicted by Gordon Bel to be about 40 percent every year between 1974 and 1984 andis now supposed to exceed 50 percent per year.