chapter 7.pdf

Computer Systems Primary Functions: Data storage and retrieval Bad performance: Sluggish I/O throughput can have a ripple effect when virtual memory is involved Amdahl’s Law: Estimate the performance improvement expect when we upgrade a system component. I/O Architectures Define: subsystem of compon...

Computer Systems Primary Functions: Data storage and retrieval Bad performance: Sluggish I/O throughput can have a ripple effect when virtual memory is involved Amdahl’s Law: Estimate the performance improvement expect when we upgrade a system component. I/O Architectures Define: subsystem of components that moves coded data between external devices and a host system. include: o Blocks of main memory that are devoted to I/O functions. o Buses that move data into and out of the system. o Control modules in the host and in peripheral devices o Interfaces to external components such as keyboards and disks. o Cabling or communications links between the host system and its peripherals. Controlled in five general ways: Programmed I/O InterruptDriven I/O MemoryMapped I/O Direct Memory Access (DMA) Channel I/O Reserves a register for each I/O device. Continuously polls the registers to detect data arrival. allows the CPU to do other things until I/O is requested. Each device connects its interrupt line to the interrupt controller. The controller signals the CPU when any of the interrupt lines are asserted. Interrupt Vectors: o set of addresses that determine executed code whenever an interrupt occurs o stored in low memory. The system state is saved before the interrupt service routine is executed and is restored afterward. Shares memory addresses between I/O devices and program memory. Devices and main memory share the same address space. Each I/O device has its own reserved block of memory. Appears as a memory access from the CPU’s perspective. Same instructions are used to move data to and from both I/O and memory. In small systems, low-level data transfer details are offloaded to I/O controllers built into the devices. offloads I/O processing to a special-purpose chip DMA and the CPU share the bus. The DMA runs at a higher priority and steals memory cycles from the CPU. I/O buses: operate asynchronously + Requests for bus access. Bus control lines: activate the devices + raise error signals + reset devices. Bus width is determined by the number of data lines. A bus clock coordinates activities and provides bit cell boundaries. uses in very large systems. consists of one or more I/O processors (IOPs) that control various channel paths. Slower devices (terminals and printers) combined (multiplexed) into a single faster channel. On IBM mainframes, multiplexed channels are called multiplexor channels, the faster ones are called selector channels. distinguished from DMA by the intelligence of the IOPs. The IOP negotiates protocols, issues device commands, translates storage coding to memory coding, and can transfer entire files or groups of files independent of the host CPU. The host has only to create the program instructions for the I/O operation and tell the IOP where to find them. Character I/O devices process one byte (or character) at a time. Examples include modems, keyboards, and mice. Keyboards are usually connected through an interrupt-driven I/O system. Block I/O devices handle bytes in groups. Most mass storage devices (disk and tape) are block I/O devices. Block I/O systems are most efficiently connected through DMA or channel I/O. Data Transmission Modes: Bytes can be conveyed from one point to another parallel data sending their encoding signals simultaneously transmission the interface requires one conductor for each bit. Parallel cables are fatter than serial cables. serial data sending them one bit at a time in serial data transmission transmission interfaces Require fewer conductors. Are less susceptible to attenuation. Can transmit data farther and faster. interfaces are suitable for time-sensitive (isochronous) data such as voice and video. Disk Technology Magnetic disks offer large amounts of durable storage that can be accessed quickly. Disk drives are called random (or direct) access storage devices, because blocks of data can be accessed according to their location on the disk. This term was coined when all other durable storage (e.g., tape) was sequential. Understanding Hard Disk Drives (HDDs): Components: Platters: mounted on spindles. Tracks: numbered from the outside edge, starting with zero Read/Write Heads: mounted on a comb that swings radially to read the disk. rotating disk forms a logical cylinder beneath the read/write heads. Data blocks are addressed by their cylinder, surface, and sector. Advantages: Low cost. Disadvantages: Slow compared to main memory. Fragile Moving parts wear out Solid State Drives (SSDs) - The Alternative: Reduced memory cost led to the widespread adoption of SSDs. Computers see SSDs as regular disk drives, but they use flash memory (like USB sticks) for data storage. Flash memory is also found in memory sticks and MP3 players. Performance Factors of HDDs: Seek Time: How long it takes the arm to move to the desired cylinder. Rotational Delay: How long it takes for the right sector to spin under the read/write head. Access Time: Seek Time + Rotational Delay. Transfer Rate: How fast data can be read from the disk once positioned. Average Latency: How long it takes on average for a sector to spin under the head (depends on rotation speed). Mean Time To Failure (MTTF): is an estimate, not a guarantee of lifespan. determined value often calculated experimentally doesn’t tell us much about the actual expected life of the disk Design life is a more realistic measure of how long an HDD can last. Understanding Solid State Drives (SSDs) Performance: SSDs are significantly faster than HDDs in access time and transfer rates (typically 100 times faster). However, they are still slower than onboard RAM by a factor of around 100,000. These numbers can vary depending on the manufacturer and interface used. Data Storage: Unlike RAM, SSDs are block-addressable, similar to hard disk drives. The duty cycle of flash is between 30,000 and 1,000,000 updates to a block. Updates are spread over the entire medium through wear leveling to prolong the life of the SSD. Comparison to HDDs: Many performance metrics used for HDDs are also applicable to SSDs. SSDs eliminate the need for metrics related to spinning components, such as rotational delay. Enterprise-Grade SSDs: highest levels of performance and reliability. Onboard cache memories use capacitors to back up data during power failures, allowing time to commit pending writes. Standardization: The Joint Electron Devices Engineering Council (JEDEC) sets standards for measuring SSD performance and reliability. Key metrics include: o Unrecoverable Bit Error Ratio (UBER): Measures data reliability. o Terabytes Written (TBW): Indicates disk endurance (service life). Optical Disks Advantages: High storage capacity at a low cost. Varieties like CD-ROM, DVD, and WORM COLD (Computer Output Laser Disk) document output on optical disk rather than on paper for large installations. Estimated lifespan of 100 years, significantly exceeding other media. CD-ROM Technology: Complex logical data format compared to magnetic disks Separate formats for data and music. Two levels of error correction for data format (limits data capacity to 650MB, allows 742MB for music). DVD Technology: Considered as a higher-density version of CD (quad-density). Varieties include single/double-sided and single/double-layered discs. CD-ROM can hold at most 650MB of data, DVDs can hold as much as 17GB. Achieves higher density through a shorter wavelength laser, allowing tighter packing of data. DVDs offer greater storage capacity than CDs Light with a shorter wavelength allows for reading and writing data at higher densities compared to longer wavelengths. Cost-effective blue-violet lasers (wavelength of 405 nm) shorter than either red (750 nm) or orange (650 nm). Blu-Ray discs utilize this shorter wavelength laser technology to achieve significantly higher data density compared to red or orange lasers used in CDs and DVDs. data centers for long-term data archiving and retrieval. Two main contenders in this space are: o Sony's Professional Disk for Data (PDD) offering 23GB per disc. o Plasmon's Ultra Density Optical (UDO) with a capacity of up to 30GB per disc. Blu-Ray Disc Technology Blu-Ray win dominance over HD-CD, due to the influence of Sony. HD-CDs offered backward compatibility with DVDs; they had lower storage capacity. Developed by a consortium including Sony, Samsung, and Pioneer, o Blu-Ray discs boast a single-layer capacity of 25GB, o with options for stacking multiple layers (up to six layers) for even greater storage. o Currently, only double-layer discs are available for home use. Magnetic Tape Storage - From Analog to High-Capacity Digital First-generation Magnetic Tape: capacity (under 11MB) + wide analog recording tape + 9 vertical tracks. Evolution to High-Capacity Digital Tapes: Modern magnetic tapes are digital and provide storage capacities in gigabytes. Two main recording methods are used: serpentine and helical scan, o differentiated by how the read/write head interacts with the tape. Serpentine: used in Digital Linear Tape (DLT) and Quarter Inch Cartridge (QIC) tape systems Helical Scan Recording: Digital Audio Tape (DAT) systems utilize helical scan recording Linear Tape-Open (LTO) standard In 1997, a collaborative by HP, IBM, and Seagate. LTO is a linear digital tape format with openly available specifications. It has seen refinement through generations (currently at Gen 5, released in 2010). Without compression, Gen 5 tapes support transfer rates of 208MB per second and hold 1.4TB of data. LTO offers multiple error correction levels for superior reliability, addressing the perception of tape as an error-prone medium. RAID (Redundant Array of Independent Disks): o address problems related to disk reliability, cost, and performance. o data is distributed across multiple disks, with additional disks added to provide error correction (redundancy). o taxonomy that has persisted for a quarter of a century, despite attempts to redefine it. Raid 0 1 2 3 Meaning Drive spinning Data is written in blocks across the entire array disk mirroring Two matched sets of disks contain the same data. set of data drives, Hamming code drives: error correction for the data drives stripes bits across a set of data drives Advantages improve performance but no redundancy 100% redundancy, and good performance personal systems. Disadvantages low reliability Cost performance is poor cost is relatively high. not for commercial provides a separate disk for parity. Parity is the XOR of the data bits. 4 5 adding parity disks to RAID 0. Data is written in blocks across the data disks, parity block is written to the redundant drive. RAID 4 would be feasible if all record blocks were the same size. RAID 4 + distributed parity. commercial systems some accesses can be serviced concurrently, giving good performance and high reliability 6 two levels of error protection over striped data: Reed-Soloman and parity. It can tolerate the loss of two disks. write-intensive, but highly fault-tolerant. Double Parity RAID (RAID DP): Employs overlapping parity blocks for linearly independent parity functions. Tolerates the loss of two disks. Better performance than RAID 6 due to simple parity functions. known as EVENODD, diagonal parity RAID, RAID 5DP, advanced data guarding RAID (RAID ADG), and— erroneously—RAID 6. Large Systems: Employ various RAID levels based on data criticality. Disk arrays for program workspace (e.g., file sorting) don’t require high fault tolerance. Critical, high-throughput files benefit from combining RAID 0 with RAID 1 (RAID 10). RAID 50 combines striping and distributed parity for fault tolerance and capacity. Higher RAID levels do not necessarily mean “better” RAID levels. It all depends upon the needs of the applications that use the disks. Advances in Magnetic Disk Storage: Technology has defied efforts to define the ultimate upper limit for magnetic disk storage. In the 1970s, the upper limit was thought to be around 2MB/in². Today’s disks commonly support 20GB/in². Improvements have occurred in several technologies, including o materials science, o magneto-optical recording heads, and o error-correcting codes. As data densities increase, bit cells consist of proportionately fewer magnetic grains. The superparamagnetic limit is the point where there are too few grains to hold a value, risking spontaneous changes from 1 to 0 or vice versa. Even if this limit is wrong by orders of magnitude, the greatest gains in magnetic storage have likely already been realized Future Data Storage Technologies: Exponential gains in data storage will likely come from entirely new technologies. Research is ongoing to find suitable replacements for magnetic disks. Interesting technologies being explored include: o Carbon nanotubes o Memristors biological data storage systems Holographic storage combine organic compounds such as proteins or oils with inorganic (magnetizable) substances. Early prototypes have encouraged the expectation that densities of 1Tb/in2 are attainable. Of course, the ultimate biological data storage medium is DNA. Trillions of messages can be stored in a tiny strand of DNA. Practical DNA-based data storage is most likely decades away. Uses laser beams to etch a three-dimensional hologram onto a polymer medium. Micro-ElectroMechanical Storage (MEMS) o ▪ o 2. o ▪ ▪ o ▪ o ▪ o ▪ 3. o o ▪ ▪ o o o 4. o Data retrieval involves passing a reference beam through the hologram to reproduce the original coded object beam. Tremendous data densities are possible due to the three-dimensional nature. Experimental systems have achieved over 30GB/in² with transfer rates around 1GBps. Content addressable, eliminating the need for a file directory. Major challenge: Finding an inexpensive, stable, and rewritable holographic medium. IBM’s Millipede is a promising MEMS device. Prototypes have achieved densities of 100GB/in², with 1Tb/in² Millipede consists of thousands of cantilevers that record a binary 1 by pressing a heated tip into a polymer substrate. CNTs (carbon nanotubes) can act as switches, opening and closing to store bits. Memristors combine resistor properties with memory, allowing controlled resistance for data storage. Amdahl’s Law quantifies the impact of I/O on performance. Components of I/O systems include memory blocks, cabling, control circuitry, interfaces, and media. I/O control methods: Programmed I/O + Interrupt-based I/O + DMA (Direct Memory Access) + Channel I/O Buses require control lines, a clock, and data lines. Timing diagrams specify operational details. Storage Technologies: Magnetic Disk: Principal form of durable storage. Performance metrics: seek time, rotational delay, and reliability estimates. Enterprise SSDs (Solid State Drives): Save energy and provide improved data access for government and industry. Optical Disks: Long-term storage for large data amounts, although access is slow. Magnetic Tape: Widely used as an archival medium. RAID (Redundant Array of Independent Disks): RAID improves disk system performance and reliability. Common RAID levels: RAID 3: Stripes data with dedicated parity disk. RAID 5: Distributed parity for concurrent access. RAID 6: Protects against dual disk failure using Reed-Solomon and parity. RAID DP (Double Parity RAID): Better performance than RAID 6. New technologies (biological, holographic, CNT, memristor, mechanical) may replace magnetic disks in the future. Data Retrieval Challenge: Locating data after it’s stored can be the hardest part of data storage.