G22.2233 L13 Disks, RAIDs, and I/O Systems. 1 Banikazemi, NYU, 2007CS G22.2233 Computer Systems Design Spring 2007Lecture 13: Disks, RAIDs, and I/O SystemsMohammad Banikazemi[Slides from Prof. Mary Jane Irwin, PSU Adapted fromComputer Organization and Design, Patterson & Hennessy, © 2005, UCB]G22.2233 L13 Disks, RAIDs, and I/O Systems. 2 Banikazemi, NYU, 2007Review: Major Components of a ComputerProcessorControlDatapathMemoryDevicesInputOutputCacheMain MemorySecondary Memory(Disk)G22.2233 L13 Disks, RAIDs, and I/O Systems. 3 Banikazemi, NYU, 2007Magnetic Disk PurposeO Long term, nonvolatile storageO Lowest level in the memory hierarchy- slow, large, inexpensiveGeneral structureO A rotating platter coated with a magnetic surfaceO A moveable read/write head to access the information on the disk Typical numbersO 1 to 4 (1 or 2 surface) platters per disk of 1” to 5.25” in diameter (3.5” dominate in 2004)O Rotational speeds of 5,400 to 15,000 RPMO 10,000 to 50,000 tracks per surface- cylinder - all the tracks under the head at a given point on all surfacesO 100 to 500 sectors per track- the smallest unit that can be read/written (typically 512B)TrackSectorG22.2233 L13 Disks, RAIDs, and I/O Systems. 4 Banikazemi, NYU, 2007Magnetic Disk Characteristic Disk read/write components1. Seek time: position the head over the proper track (3 to 14 ms avg)- due to locality of disk references the actual average seek time may be only 25% to 33% of the advertised number2. Rotational latency: wait for the desired sector to rotate under the head (½ of 1/RPM converted to ms)- 0.5/5400RPM = 5.6ms to 0.5/15000RPM = 2.0ms3. Transfer time: transfer a block of bits (one or more sectors) under the head to the disk controller’s cache (30 to 80 MB/s aretypical disk transfer rates)- the disk controller’s “cache” takes advantage of spatial locality in disk accesses– cache transfer rates are much faster (e.g., 320 MB/s)4. Controller time: the overhead the disk controller imposes in performing a disk I/O access (typically < .2 ms)SectorTrackCylinderHeadPlatterController+CacheG22.2233 L13 Disks, RAIDs, and I/O Systems. 5 Banikazemi, NYU, 2007Typical Disk Access Time The average time to read or write a 512B sector for a disk rotating at 10,000RPM with average seek time of 6ms, a 50MB/sec transfer rate, and a 0.2ms controller overheadIf the measured average seek time is 25% of the advertised average seek time, then The rotational latency is usually the largest component of the access timeG22.2233 L13 Disks, RAIDs, and I/O Systems. 6 Banikazemi, NYU, 2007Typical Disk Access Time If the measured average seek time is 25% of the advertised average seek time, thenAvg disk read/write = 6.0ms + 0.5/(10000RPM/(60sec/minute) )+ 0.5KB/(50MB/sec) + 0.2ms = 6.0 + 3.0 + 0.01 + 0.2 = 9.21msAvg disk read/write = 1.5 + 3.0 + 0.01 + 0.2 = 4.71ms The average time to read or write a 512B sector for a disk rotating at 10,000RPM with average seek time of 6ms, a 50MB/sec transfer rate, and a 0.2ms controller overhead The rotational latency is usually the largest component of the access timeG22.2233 L13 Disks, RAIDs, and I/O Systems. 7 Banikazemi, NYU, 2007Magnetic Disk Examples (www.seagate.com)1.41612/8/191”x4”x5.8”600,0008.5r-9.5w1.0r-1.2w32-587,20042003.5Seagate ST321.5r-2.0w0.2r-0.4wMinimum seek (ms)3457-86Transfer rate (MB/sec)0.4”x2.7”x3.9”1”x4”x5.8”Dimensions (inches)330,0001,200,000MTTF (hours@25oC)0.21.9Weight (pounds)174GB/watt2.4/1/0.420?/12/-Power: op/idle/sb (watts)103GB/cu.inch12r-14w3.6r-3.9wAverage seek (ms)5,40015,000Rotation speed (RPM)28# of surfaces (heads)4073.4Capacity (GB)2.53.5Disk diameter (inches)Seagate ST94Seagate ST37CharacteristicG22.2233 L13 Disks, RAIDs, and I/O Systems. 8 Banikazemi, NYU, 2007Disk Latency & Bandwidth Milestones Disk latency is one average seek time plus the rotational latency. Disk bandwidth is the peak transfer time of formatted data from the media (not from the cache).5.78.812.717.148.3Latency (msec)8624940.6Bandwidth (MB/s)SCSISCSISCSISCSIST-412Interface2.53.03.55.255.25Diameter (inches)73.49.14.31.40.03Capacity (Gbytes)20031998199419901983Year1500010000720054003600RSpeed (RPM)SG ST37SG ST39SG ST15SG ST41CDC WrenPatterson, CACM Vol 47, #10, 2004G22.2233 L13 Disks, RAIDs, and I/O Systems. 9 Banikazemi, NYU, 2007Latency & Bandwidth Improvements In the time that the disk bandwidth doubles the latencyimproves by a factor of only 1.2 to 1.40204060801001983 1990 1994 1998 2003Year of IntroductionBandwidth (MB/s)Latency (msec)G22.2233 L13 Disks, RAIDs, and I/O Systems. 10 Banikazemi, NYU, 2007Aside: Media Bandwidth/Latency Demands Bandwidth requirementsO High quality video- Digital data = (30 frames/s) × (640 x 480 pixels) × (24-b color/pixel) = 221 Mb/s (27.625 MB/s)O High quality audio- Digital data = (44,100 audio samples/s) × (16-b audio samples) ×(2 audio channels for stereo) = 1.4 Mb/s (0.175 MB/s)O Compression reduces the bandwidth requirements considerably Latency issuesO How sensitive is your eye (ear) to variations in video (audio) rates?O How can you ensure a constant rate of delivery?O How important is synchronizing the audio and video streams?- 15 to 20 ms early to 30 to 40 ms late is tolerableG22.2233 L13 Disks, RAIDs, and I/O Systems. 11 Banikazemi, NYU, 2007Dependability, Reliability, Availability Reliability – measured by the mean time to failure(MTTF). Service interruption is measured by mean time to repair (MTTR) Availability – a measure of service accomplishmentAvailability = MTTF/(MTTF + MTTR) To increase MTTF, either improve the quality of the components or design the system to continue operating in the presence of faulty components1. Fault avoidance: preventing fault occurrence by construction2. Fault tolerance: using redundancy to correct or bypass faulty components (hardware)O Fault detection versus fault correctionO Permanent faults versus transient faultsG22.2233 L13 Disks, RAIDs, and I/O Systems. 12 Banikazemi, NYU, 2007RAIDs: Disk Arrays Arrays of small and inexpensive disksO Increase potential throughput by having many disk drives- Data is spread over multiple disk- Multiple accesses are made to several disks at a timeReliability is lower than a single disk But availability can be improved by adding redundant disks (RAID)O Lost information can be reconstructed from redundant informationO
View Full Document