CPSC 614:Graduate Computer Architecture Prof. Lawrence Rauchwerger I/O Introduction: Storage Devices & RAIDMotivation: Who Cares About I/O?Big Picture: Who cares about CPUs?I/O SystemsStorage Technology DriversOutlineDisk Device TerminologyPhoto of Disk Head, Arm, ActuatorDisk Device PerformanceSlide 10Data Rate: Inner vs. Outer TracksDevices: Magnetic DisksDisk Performance Model /TrendsState of the Art: Barracuda 180Disk Performance Example (will fix later)Areal DensitySlide 18MBits per square inch: DRAM as % of Disk over timeHistorical PerspectiveDisk HistorySlide 221 inch disk drive!Disk Characteristics in 2000Slide 25Slide 26Slide 27Fallacy: Use Data Sheet “Average Seek” TimeFallacy: Use Data Sheet Transfer RateDisk Performance ExampleFuture Disk Size and PerformanceWhat about FLASHTape vs. DiskCurrent Drawbacks to TapeAutomated Cartridge System: StorageTek Powderhorn 9310Library vs. StorageWhither tape?Use Arrays of Small Disks?Advantages of Small Formfactor Disk DrivesReplace Small Number of Large Disks with Large Number of Small Disks! (1988 Disks)Array ReliabilityRedundant Arrays of (Inexpensive) DisksRedundant Arrays of Inexpensive Disks RAID 1: Disk Mirroring/ShadowingRedundant Array of Inexpensive Disks RAID 3: Parity DiskRAID 3Inspiration for RAID 4Redundant Arrays of Inexpensive Disks RAID 4: High I/O Rate ParityInspiration for RAID 5Redundant Arrays of Inexpensive Disks RAID 5: High I/O Rate Interleaved ParityProblems of Disk Arrays: Small WritesSystem Availability: Orthogonal RAIDsSystem-Level AvailabilityBerkeley History: RAID-ISummary: RAID Techniques: Goal was performance, popularity due to reliability of storageSummary StorageCPSC 614:Graduate Computer Architecture Prof. Lawrence Rauchwerger I/O Introduction: Storage Devices & RAIDBased on lectures byProf. David CullerProf. David PattersonUC BerkeleyMotivation: Who Cares About I/O?•CPU Performance: 60% per year•I/O system performance limited by mechanical delays (disk I/O)< 10% per year (IO per sec)•Amdahl's Law: system speed-up limited by the slowest part!10% IO & 10x CPU => 5x Performance (lose 50%)10% IO & 100x CPU => 10x Performance (lose 90%)• I/O bottleneck: Diminishing fraction of time in CPUDiminishing value of faster CPUsBig Picture: Who cares about CPUs?•Why still important to keep CPUs busy vs. IO devices ("CPU time"), as CPUs not costly?–Moore's Law leads to both large, fast CPUs but also to very small, cheap CPUs–2001 Hypothesis: 600 MHz PC is fast enough for Office Tools?–PC slowdown since fast enough unless games, new apps?•People care more about about storing information and communicating information than calculating–"Information Technology" vs. "Computer Science"–1960s and 1980s: Computing Revolution–1990s and 2000s: Information AgeI/O SystemsProcessorCacheMemory - I/O BusMainMemoryI/OControllerDisk DiskI/OControllerI/OControllerGraphicsNetworkinterruptsinterruptsStorage Technology Drivers•Driven by the prevailing computing paradigm–1950s: migration from batch to on-line processing–1990s: migration to ubiquitous computing»computers in phones, books, cars, video cameras, …»nationwide fiber optical network with wireless tails•Effects on storage industry:–Embedded storage»smaller, cheaper, more reliable, lower power–Data utilities»high capacity, hierarchically managed storageOutline•Disk Basics•Disk History•Disk options in 2000•Disk fallacies and performance•FLASH•Tapes•RAIDDisk Device Terminology•Several platters, with information recorded magnetically on both surfaces (usually)•Actuator moves head (end of arm,1/surface) over track (“seek”), select surface, wait for sector rotate under head, then read or write– “Cylinder”: all tracks under heads •Bits recorded in tracks, which in turn divided into sectors (e.g., 512 Bytes)PlatterOuterTrackInnerTrackSectorActuatorHeadArmPhoto of Disk Head, Arm, ActuatorActuatorArmHeadPlatters (12){SpindleDisk Device PerformancePlatterArmActuatorHeadSectorInnerTrackOuterTrack•Disk Latency = Seek Time + Rotation Time + Transfer Time + Controller Overhead•Seek Time? depends no. tracks move arm, seek speed of disk•Rotation Time? depends on speed disk rotates, how far sector is from head •Transfer Time? depends on data rate (bandwidth) of disk (bit density), size of requestControllerSpindleDisk Device Performance•Average distance sector from head?•1/2 time of a rotation–10000 Revolutions Per Minute 166.67 Rev/sec–1 revolution = 1/ 166.67 sec  6.00 milliseconds–1/2 rotation (revolution)  3.00 ms•Average no. tracks move arm?–Sum all possible seek distances from all possible tracks / # possible»Assumes average seek distance is random–Disk industry standard benchmarkData Rate: Inner vs. Outer Tracks •To keep things simple, orginally kept same number of sectors per track–Since outer track longer, lower bits per inch•Competition  decided to keep BPI the same for all tracks (“constant bit density”) More capacity per disk More of sectors per track towards edge Since disk spins at constant speed, outer tracks have faster data rate•Bandwidth outer track 1.7X inner track!–Inner track highest density, outer track lowest, so not really constant–2.1X length of track outer / inner, 1.7X bits outer / innerDevices: Magnetic DisksSectorTrackCylinderHeadPlatter•Purpose:– Long-term, nonvolatile storage– Large, inexpensive, slow level in the storage hierarchy•Characteristics:– Seek Time (~8 ms avg)»positional latency»rotational latency• Transfer rate– 10-40 MByte/sec–Blocks• Capacity–Gigabytes–Quadruples every 2 years (aerodynamics)7200 RPM = 120 RPS => 8 ms per rev ave rot. latency = 4 ms128 sectors per track => 0.25 ms per sector1 KB per sector => 16 MB / sResponse time = Queue + Controller + Seek + Rot + XferService timeDisk Performance Model /Trends• Capacity+ 100%/year (2X / 1.0 yrs)•Transfer rate (BW)+ 40%/year (2X / 2.0 yrs)•Rotation + Seek time– 8%/ year (1/2 in 10 yrs)•MB/$> 100%/year (2X / 1.0 yrs)Fewer chips + areal densityState of the Art: Barracuda 180–181.6 GB, 3.5 inch disk–12 platters, 24 surfaces–24,247 cylinders–7,200 RPM; (4.2 ms avg. latency)–7.4/8.2 ms avg. seek (r/w)–64 to 35 MB/s (internal)–0.1 ms controller time–10.3 watts (idle)source: www.seagate.comLatency = Queuing Time + Controller time +Seek Time + Rotation Time + Size / Bandwidthper accessper