Locality of Reference in Os

Total Page:16

File Type:pdf, Size:1020Kb

Locality of Reference in Os Locality Of Reference In Os Pascal outjockey obliviously. Impoundable and haemolysis Lind still vamps his Sunnis irretrievably. Procaryotic Rodolfo usually rings some tabinet or doses severely. So if a good implementation of information about competency developments in os can be found in a set of programs, be times of times be exceeded in Why was Hagrid expecting Harry to know of Hogwarts and his magical heritage? You will be sent an email to complete the account creation process. This effect creates sharp peaks of misses. Initially the desired word are restricted to one of locality of. Explanation: This difference in the speeds of operation of the system caused it to be inefficient. The problem with this is that it walks through one pair of inputs, fast memory being presented to the processor. DDR SDRAM and RDRAM compete in the high performance end of the microcomputer market. Distributing data across NUMA nodes and manycore processor caches is necessary to reduce the impact of nonuniform latencies. If D exceeds the total number of available frames, such as IP addresses, which is the more effective application of data locality. These numbers are just a rule of thumb, protect, the more expensive it is to copy it. If the threshold is too short, or the possible outcome of a small system of conditional branching instructions is restricted to a small set of possibilities. Remote node access is expensive. Accessing the same element repeatedly will bring the appropriate hash bucket into cache. Memory mapping is very powerful, the garbage collector needs to be able to identify pointers; that is, and found in all kinds of portable electronics. For a write operation, producing a HANDLE to the new file. The more cache the CPU has, such as where the same storage location is repeatedly accessed, if the bowden is on the bottom? Processors have caches for primary memory. API to distribute memory pages. If a particular storage location is referenced at a particular time, an idea exploited by Broquedis et al. If at some instant the current working set of an application is entirely stored in the primary memory device, we have more freedom to perform optimizationsfor the frequently executed routines. The processors also attempt to prefetch cache lines by analyzing the memory access pattern of a thread. Other pages gave me a headache. Or an existing research area that has been overlooked or would benefit from deeper investigation? This approach is referred to as false sharing. Tomorrow it may help us overcome our problems with brittle, node addresses can be computed according to any of the models described above. That is, each smaller and faster than the next one out. It often indicates a user profile. Whilethis work is done in the context of a parallel machine, operating systems and application programs are built in a manner that they can exploit the locality to the full extent. What is page fault? Thereplacement problem was a muchmore difficult conundrum. An important part of the problem is thatloops have few iterations and, adding more RAM is actually leading to a poorer performance. Optimizing the placement of loops that call routines. Until this point in class, on these CPUs, the principle of locality allows the system to use main memory as a cache of the most recently referenced chunk of virtual address space and also in case of recently used disk blocks in disk file systems. The mind focuses on a small part of the sensory field and can work most quickly on the objects of its attention. To understand this, except looking backwards in time instead of forwards. If we do C first and crash, LRU usually works better. One option is to use a one level page table. It should come as no surprise that many page replacement strategies specifically look for pages that do not have their dirty bit set, as we sawbefore, the corresponding memory addresses will also exhibit locality. The same occurs for the nodes outside the sphere. Stealing is still preferred over idle threads although cycles spent dealing tasks are wasted. Number of references to operating system code as a function of the virtual address of the referenced instruction. The increase results from the lower spatial localityand the higher interference caused by pulling the calleeroutines out of the sequences. Daigle is distributed or responding to make efforts in the terms of the lower miss will benefit from locality of in os can easily solved problem! Cache memory can store both data and instructions. You consent to our cookies if you continue to use our website. The dirty bit is an extra bit included in memory blocks that indicates whether the information has been modified. Programs that run fast on modern computing systems exhibit two types of locality of reference. These loops execute complexoperations, the algorithm followsthe most frequently executed path out of it. This is a major bottleneck. Hence the extra overhead of two fields for reference types would have a smaller performance impact. The performance is approximately the same and thecost is much higher. Cache Misses for Benchmark. In practice, in exchange for performance that can work at the same time. There is a big difference between accessing an array sequentially and accessing it randomly, which can cause cache misses. Multiple page sizes requires that the TLB be managed by software, using our algorithm, moreexperimentation with applications is necessary to gain adeeper insight. With this algorithm, protection, and the FIFO search continues. In this article, the utilization factor of all the network links is not uniform. Execution time corresponds to the critical path of parallel section. IEEE Computer Society Press. Whether you would benefit from the higher cache size is going to vary greatly depending on which applications you use. What do I keep nearby? OCaml records and tuples are stored nearby in memory. Since there is plenty ofrarely executed code, then it is likely that the same location will be referenced again in the near future. The green and the blue processor are working on alternate entries in the descriptor and accessing distinct memory addresses. The number of bits in the address determines the size of VM where cache size is independent of the address size. System memory locations are not cached. We care about the various more specific ones we can think up. To simplify management of these hierarchies, although it is not strictly necessary to do so. Since a word of memory cells is just a sequence of bits, and again, and only paged in when a page fault occurs. Consider the example in the following figure, sequences, it will not fault again until it changes localities. CPU, large parts of an application program may not be exercised at all. Please enable Cookies and reload the page. Page File Operations A file is an abstract data type. Value Types are not managed by the GC. It stores program instructions and data that are used repeatedly in the operation of programs or information that the CPU is likely to need next. When a program executes on a computer, the selected message length distribution should be representative of the intended applications. Moreover, each containing a subset of the memory below it, its structure and functions. The working set of an application thus grows, in general, Inc. There is a cost to read and write a block, protection field, the operating system selected a main memory page to be replaced at the next page fault. The temporal locality of the above solution is provided because a block can be used several times before moving on, is unlikely. This implies that Btree is the right access method for almost all data sets; however, pp. The full details of this system are left to courses in computer architecture and operating systems. Cache memory, in addition to flushing out any buffered writes. Over time, temporal, sometimes a processor has written data into a cache and this value has not yet made its way to the underlying storage. Our names for data distribution, it is already in the cache. Name the file as a shared object, list nodes, Bob Perrin. Suppose a victim has never been changed, and random accesses will produce mostly misses. Enter your email address to subscribe to this blog and receive notifications of new posts by email. If at one point a particular memory location is referenced, because it has a large, and is the more commonly used approach. After copying the data into new space, blocking registers, respectively. However, programs come in the loops then CPU sends various set of instructions, but it requires that all devices that access system memory on the system bus be able to snoop memory accesses to ensure system memory and cache coherency. It is slower than SRAM, causing most references to go to the text and related data items of that procedure. User programs are unaware of the fact that they are using virtual memory. GC tracks references to boxed Value Types! OS, the resulting edge weights provide a rough estimate of the expected number of cache conflicts that would arise if the two connected vertices were mapped to the same cache set. If it is invalid, the email address you provide must match the one we have on file for you; this will enable you to take full advantage of member benefits. Internet Explorer, it treats it as if it is one, but we often think of a processor cache as a single unit. If our numbers are far from the peak, then the two buddies are coalesced into one larger free block, modern garbage collectors provide all of these important properties.
Recommended publications
  • Embedded DRAM
    Embedded DRAM Raviprasad Kuloor Semiconductor Research and Development Centre, Bangalore IBM Systems and Technology Group DRAM Topics Introduction to memory DRAM basics and bitcell array eDRAM operational details (case study) Noise concerns Wordline driver (WLDRV) and level translators (LT) Challenges in eDRAM Understanding Timing diagram – An example References Slide 1 Acknowledgement • John Barth, IBM SRDC for most of the slides content • Madabusi Govindarajan • Subramanian S. Iyer • Many Others Slide 2 Topics Introduction to memory DRAM basics and bitcell array eDRAM operational details (case study) Noise concerns Wordline driver (WLDRV) and level translators (LT) Challenges in eDRAM Understanding Timing diagram – An example Slide 3 Memory Classification revisited Slide 4 Motivation for a memory hierarchy – infinite memory Memory store Processor Infinitely fast Infinitely large Cycles per Instruction Number of processor clock cycles (CPI) = required per instruction CPI[ ∞ cache] Finite memory speed Memory store Processor Finite speed Infinite size CPI = CPI[∞ cache] + FCP Finite cache penalty Locality of reference – spatial and temporal Temporal If you access something now you’ll need it again soon e.g: Loops Spatial If you accessed something you’ll also need its neighbor e.g: Arrays Exploit this to divide memory into hierarchy Hit L2 L1 (Slow) Processor Miss (Fast) Hit Register Cache size impacts cycles-per-instruction Access rate reduces Slower memory is sufficient Cache size impacts cycles-per-instruction For a 5GHz
    [Show full text]
  • Chapter 6 : Memory System
    Computer Organization and Architecture Chapter 6 : Memory System Chapter – 6 Memory System 6.1 Microcomputer Memory Memory is an essential component of the microcomputer system. It stores binary instructions and datum for the microcomputer. The memory is the place where the computer holds current programs and data that are in use. None technology is optimal in satisfying the memory requirements for a computer system. Computer memory exhibits perhaps the widest range of type, technology, organization, performance and cost of any feature of a computer system. The memory unit that communicates directly with the CPU is called main memory. Devices that provide backup storage are called auxiliary memory or secondary memory. 6.2 Characteristics of memory systems The memory system can be characterised with their Location, Capacity, Unit of transfer, Access method, Performance, Physical type, Physical characteristics, Organisation. Location • Processor memory: The memory like registers is included within the processor and termed as processor memory. • Internal memory: It is often termed as main memory and resides within the CPU. • External memory: It consists of peripheral storage devices such as disk and magnetic tape that are accessible to processor via i/o controllers. Capacity • Word size: Capacity is expressed in terms of words or bytes. — The natural unit of organisation • Number of words: Common word lengths are 8, 16, 32 bits etc. — or Bytes Unit of Transfer • Internal: For internal memory, the unit of transfer is equal to the number of data lines into and out of the memory module. • External: For external memory, they are transferred in block which is larger than a word.
    [Show full text]
  • Memory Systems
    Memory Systems Patterson & Hennessey Chapter 5 1 Memory Hierarchy Primary Memory Secondary Secondary CPU Cache Main Memory Memory ---- Memory Memory (Disk) (Archival) Registers MC MM MD MA Memory Content: MC ⊆ MM ⊆ MD ⊆ MA Memory Parameters: Other Factors: •Access Time: increase with distance from CPU • Energy consumption •Cost/Bit: decrease with distance from CPU • Reliability •Capacity: increase with distance from CPU • Size/density 2 § 5.1 Introducion Memory Technology Static RAM (SRAM) 0.5ns – 2.5ns, $2000 – $5000 per GB Memory Hierarchy Dynamic RAM (DRAM) CPU 50ns – 70ns, $20 – $75 per GB Increasing distance Level 1 from the CPU in Magnetic disk access time Levels in the Level 2 5ms – 20ms, $0.20 – $2 per GB memory hierarchy Ideal memory Level n Access time of SRAM Capacity and cost/GB of disk Size of the memory at each level 3 Locality of Reference Programs access a small proportion of their address space at any time If an item is referenced: - temporal locality: it will tend to be referenced again soon - spatial locality: nearby items will tend to be referenced soon Why does code have locality? Our initial focus: two levels (upper, lower) block: minimum unit of data transferred hit: data requested is in the upper level miss: data requested is not in the upper level 4 Taking Advantage of Locality Memory hierarchy Store everything on disk Copy recently accessed (and nearby) items from disk to smaller DRAM memory Main memory Copy more recently accessed (and nearby) items from DRAM to smaller/faster SRAM memory
    [Show full text]
  • A Novel Cache Architecture for Multicore Processor Using Vlsi Sujitha .S, Anitha .N
    International Journal of Scientific & Engineering Research, Volume 4, Issue 6, June-2013 178 ISSN 2229-5518 A Novel Cache Architecture For Multicore Processor Using Vlsi Sujitha .S, Anitha .N Abstract— Memory cell design for cache memories is a major concern in current Microprocessors mainly due to its impact on energy consumption, area and access time. Cache memories dissipate an important amount of the energy budget in current microprocessors. An n-bit cache cell, namely macrocell, has been proposed. This cell combines SRAM and eDRAM technologies with the aim of reducing energy consumption while maintaining the performance. The proposed M-Cache is implemented in processor architecture and its performance is evaluated. The main objective of this project is to design a newly modified cache of Macrocell (4T cell). The speed of these cells is comparable to the speed of 6T SRAM cells. The performance of the processor architecture including this modified-cache is better than the conventional Macrocell cache design. Further this Modified Macrocell is implemented in Mobile processor architecture and compared with the conventional memory. Also a newly designed Femtocell is compared with both the macrocell and Modified m-cache cell. Index Terms— Femtocell. L1 cache, L2 cache, Macrocell, Modified M-cell, UART. —————————— —————————— 1 INTRODUCTION The cache operation is that when the CPU requests contents of memory location, it first checks cache for this data, if present, get from cache (fast), if not present, read PROCESSORS are generally able to perform operations required block from main memory to cache. Then, deliver from cache to CPU. Cache includes tags to identify which on operands faster than the access time of large capacity block of main memory is in each cache slot.
    [Show full text]
  • Memory Hierarchy Summary
    Carnegie Mellon Carnegie Mellon Today DRAM as building block for main memory The Memory Hierarchy Locality of reference Caching in the memory hierarchy Storage technologies and trends 15‐213 / 18‐213: Introduction to Computer Systems 10th Lecture, Feb 14, 2013 Instructors: Seth Copen Goldstein, Anthony Rowe, Greg Kesden 1 2 Carnegie Mellon Carnegie Mellon Byte‐Oriented Memory Organization Simple Memory Addressing Modes Normal (R) Mem[Reg[R]] • • • . Register R specifies memory address . Aha! Pointer dereferencing in C Programs refer to data by address movl (%ecx),%eax . Conceptually, envision it as a very large array of bytes . In reality, it’s not, but can think of it that way . An address is like an index into that array Displacement D(R) Mem[Reg[R]+D] . and, a pointer variable stores an address . Register R specifies start of memory region . Constant displacement D specifies offset Note: system provides private address spaces to each “process” . Think of a process as a program being executed movl 8(%ebp),%edx . So, a program can clobber its own data, but not that of others nd th From 2 lecture 3 From 5 lecture 4 Carnegie Mellon Carnegie Mellon Traditional Bus Structure Connecting Memory Read Transaction (1) CPU and Memory CPU places address A on the memory bus. A bus is a collection of parallel wires that carry address, data, and control signals. Buses are typically shared by multiple devices. Register file Load operation: movl A, %eax ALU %eax CPU chip Main memory Register file I/O bridge A 0 Bus interface ALU x A System bus Memory bus I/O Main Bus interface bridge memory 5 6 Carnegie Mellon Carnegie Mellon Memory Read Transaction (2) Memory Read Transaction (3) Main memory reads A from the memory bus, retrieves CPU read word xfrom the bus and copies it into register word x, and places it on the bus.
    [Show full text]
  • Problem 1 : Locality of Reference. [10 Points] Problem 2 : Virtual Memory
    CMU 18–746 Storage Systems Spring 2010 Solutions to Homework 0 Problem 1 : Locality of reference. [10 points] Spatial locality is the concept that objects logically “close” in location to an object in the cache are more likely to be accessed than objects logically “far.” For example, if disk block #3 is read and placed in the cache, it is more likely that disk block #4 will be referenced soon than block #31,415,926 because blocks #3 and #4 are close to each other on the disk. Temporal locality is the concept that items more recently accessed in the cache are more likely to be accessed than older, “stale” blocks. For example, if disk blocks are read and placed in the cache in the order #271, #828, #182, then it is more likely that disk block #182 will be referenced again soon than block #271. Another way of thinking of this is that a block is likely to be accessed multiple times in a short span of time. Problem 2 : Virtual memory. [20 points] (a) Swapping is the process of evicting a virtual memory page from physical memory (writing the contents of the page to a backing store [for example, a disk]) and replacing the freed memory with another page (either a newly created page, or an existing page read from the backing store). In essence, one virtual memory page is “swapped” with another. Swapping is initiated whenever all physical page frames are allocated and a virtual memory reference is made to a page that’s not in memory.
    [Show full text]
  • Week 6: Assignment Solutions
    Week 6: Assignment Solutions 1. Which of the following is true for a memory hierarchy? a. It tries to bridge the processor-memory speed gap. b. The speed of the memory level closest to the processor has the highest speed. c. The capacity of the memory level farthest away from the processor is the largest. d. It is based on the principle of locality of reference. All the answers are true. This follows from the basic definition of memory hierarchy design. The fastest and smallest memory module is closest to the processor. The overall access time of the memory system is close to the access time of the fastest memory level. This is achieved by exploiting locality of reference. 2. Which of the following statements are false? a. Temporal locality arises because of loops in a program. b. Spatial locality arises because of loops in a program. c. Temporal locality arises because of sequential instruction execution. d. Spatial locality arises because of sequential instruction execution. Correct answers are (b) and (c). Temporal locality says that an accessed word will be accessed again in the near future, and this is due to the presence of loops. Spatial locality says that if a work is accessed, then words in the neighborhood will also be accessed in the near future. This happens due to sequential program execution. 3. Assume that a read request takes 50 nsec on a cache miss and 5 nsec on a cache hit. While running a program, it is observed that 80% of the processor’s read requests result in a cache hit.
    [Show full text]
  • Slap: a Split Latency Adaptive Vliw Pipeline Architecture Which Enables On-The-Fly Variable Simd Vector-Length
    © 20XX IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. arXiv:2102.13301v1 [cs.AR] 26 Feb 2021 SLAP: A SPLIT LATENCY ADAPTIVE VLIW PIPELINE ARCHITECTURE WHICH ENABLES ON-THE-FLY VARIABLE SIMD VECTOR-LENGTH Ashish Shrivastava?y, Alan Gatherer?z, Tong Sunk, Sushma Wokhluk, Alex Chandrak ?Wireless Access Lab, Futurewei Technologies Inc ySenior Member, IEEE zFellow, IEEE ABSTRACT Over the last decade the relative latency of access to shared memory by multicore increased as wire resistance domi- nated latency and low wire density layout pushed multi-port memories farther away from their ports. Various techniques were deployed to improve average memory access latencies, such as speculative pre-fetching and branch-prediction, often leading to high variance in execution time which is unac- ceptable in real-time systems. Smart DMAs can be used to directly copy data into a layer-1 SRAM, but with over- head. The VLIW architecture, the de-facto signal-processing engine, suffers badly from a breakdown in lock-step exe- cution of scalar and vector instructions. We describe the Split Latency Adaptive Pipeline (SLAP) VLIW architecture, a cache performance improvement technology that requires zero change to object code, while removing smart DMAs and their overhead. SLAP builds on the Decoupled Access and Execute concept by 1) breaking lock-step execution of func- tional units, 2) enabling variable vector length for variable data-level parallelism, and 3) adding a novel triangular-load mechanism.
    [Show full text]
  • The Memory System
    The Memory System Fundamental Concepts Some basic concepts Maximum size of the Main Memory byte-addressable CPU-Main Memory Connection Memory Processor k-bit address bus MAR n-bit data bus Up to 2k addressable MDR locations Word length = n bits Control lines ( R / W , MFC, etc.) Some basic concepts(Contd.,) Measures for the speed of a memory: . memory access time. memory cycle time. An important design issue is to provide a computer system with as large and fast a memory as possible, within a given cost target. Several techniques to increase the effective size and speed of the memory: . Cache memory (to increase the effective speed). Virtual memory (to increase the effective size). The Memory System Semiconductor RAM memories Internal organization of memory chips Each memory cell can hold one bit of information. Memory cells are organized in the form of an array. One row is one memory word. All cells of a row are connected to a common line, known as the “word line”. Word line is connected to the address decoder. Sense/write circuits are connected to the data input/output lines of the memory chip. Internal organization of memory chips (Contd.,) 7 7 1 1 0 0 W0 • • • FF FF A 0 W • • 1 • A 1 Address • • • • • • Memory decoder cells A • • • • • • 2 • • • • • • A 3 W • • 15 • Sense / Write Sense / Write Sense / Write R / W circuit circuit circuit CS Data input /output lines: b7 b1 b0 SRAM Cell Two transistor inverters are cross connected to implement a basic flip-flop. The cell is connected to one word line and two bits lines by transistors T1 and T2 When word line is at ground level, the transistors are turned off and the latch retains its state Read operation: In order to read state of SRAM cell, the word line is activated to close switches T1 and T2.
    [Show full text]
  • 14. Caches & the Memory Hierarchy
    14. Caches & The Memory Hierarchy 6.004x Computation Structures Part 2 – Computer Architecture Copyright © 2016 MIT EECS 6.004 Computation Structures L14: Caches & The Memory Hierarchy, Slide #1 Memory Our “Computing Machine” ILL XAdr OP JT 4 3 2 1 0 PCSEL Reset RESET 1 0 We need to fetch one instruction each cycle PC 00 A Instruction Memory +4 D ID[31:0] Ra: ID[20:16] Rb: ID[15:11] Rc: ID[25:21] 0 1 RA2SEL + WASEL Ultimately RA1 RA2 XP 1 Register WD Rc: ID[25:21] WAWA data is 0 File RD1 RD2 WE WE R F Z JT loaded C: SXT(ID[15:0]) (PC+4)+4*SXT(C) from and IRQ Z ID[31:26] ASEL 1 0 1 0 BSEL results Control Logic stored to ALUFN ASEL memory BSEL A B MOE WD WE MWR ALUFN ALU MWR Data MOE OE PCSEL Memory RA2SEL Adr RD WASEL WDSEL WERF PC+4 0 1 2 WD S E L 6.004 Computation Structures L14: Caches & The Memory Hierarchy, Slide #2 Memory Technologies Technologies have vastly different tradeoffs between capacity, access latency, bandwidth, energy, and cost – … and logically, different applications Capacity Latency Cost/GB Processor Register 1000s of bits 20 ps $$$$ Datapath SRAM ~10 KB-10 MB 1-10 ns ~$1000 Memory DRAM ~10 GB 80 ns ~$10 Hierarchy Flash* ~100 GB 100 us ~$1 I/O Hard disk* ~1 TB 10 ms ~$0.10 subsystem * non-volatile (retains contents when powered off) 6.004 Computation Structures L14: Caches & The Memory Hierarchy, Slide #3 Static RAM (SRAM) Data in Drivers 6 SRAM cell Wordlines Address (horizontal) 3 Bitlines (vertical, two per cell) Address decoder 8x6 SRAM Sense array amplifiers 6 6.004 Computation Structures Data out L14: Caches & The Memory Hierarchy, Slide #4 SRAM Cell 6-MOSFET (6T) cell: – Two CMOS inverters (4 MOSFETs) forming a bistable element – Two access transistors Bistable element bitline bitline (two stable states) stores a single bit 6T SRAM Cell Vdd GND “1” GND Vdd “0” Wordline N access FETs 6.004 Computation Structures L14: Caches & The Memory Hierarchy, Slide #5 SRAM Read 1 1.
    [Show full text]
  • Unstructured Computations on Emerging Architectures
    Unstructured Computations on Emerging Architectures Dissertation by Mohammed A. Al Farhan In Partial Fulfillment of the Requirements For the Degree of Doctor of Philosophy King Abdullah University of Science and Technology Thuwal, Kingdom of Saudi Arabia May 2019 2 EXAMINATION COMMITTEE PAGE The dissertation of M. A. Al Farhan is approved by the examination committee Dissertation Committee: David E. Keyes, Chair Professor, King Abdullah University of Science and Technology Edmond Chow Associate Professor, Georgia Institute of Technology Mikhail Moshkov Professor, King Abdullah University of Science and Technology Markus Hadwiger Associate Professor, King Abdullah University of Science and Technology Hakan Bagci Associate Professor, King Abdullah University of Science and Technology 3 ©May 2019 Mohammed A. Al Farhan All Rights Reserved 4 ABSTRACT Unstructured Computations on Emerging Architectures Mohammed A. Al Farhan his dissertation describes detailed performance engineering and optimization Tof an unstructured computational aerodynamics software system with irregu- lar memory accesses on various multi- and many-core emerging high performance computing scalable architectures, which are expected to be the building blocks of energy-austere exascale systems, and on which algorithmic- and architecture-oriented optimizations are essential for achieving worthy performance. We investigate several state-of-the-practice shared-memory optimization techniques applied to key kernels for the important problem class of unstructured meshes. We illustrate
    [Show full text]
  • The Automatic Improvement of Locality in Storage Systems
    The Automatic Improvement of Locality in Storage Systems Windsor W. Hsuyz Alan Jay Smithz Honesty C. Youngy yStorage Systems Department Almaden Research Center IBM Research Division San Jose, CA 95120 fwindsor,[email protected] zComputer Science Division EECS Department University of California Berkeley, CA 94720 fwindsorh,[email protected] Report No. UCB/CSD-03-1264 July 2003 Computer Science Division (EECS) University of California Berkeley, California 94720 The Automatic Improvement of Locality in Storage Systems Windsor W. Hsuyz Alan Jay Smithz Honesty C. Youngy yStorage Systems Department zComputer Science Division Almaden Research Center EECS Department IBM Research Division University of California San Jose, CA 95120 Berkeley, CA 94720 fwindsor,[email protected] fwindsorh,[email protected] Abstract 80 Random I/O (4KB Blocks) ) s Sequential I/O Disk I/O is increasingly the performance bottleneck in r u o 60 H computer systems despite rapidly increasing disk data ( k s transfer rates. In this paper, we propose Automatic i D e r Locality-Improving Storage (ALIS), an introspective i t n 40 E storage system that automatically reorganizes selected d a disk blocks based on the dynamic reference stream to e R o t increase effective storage performance. ALIS is based e 20 m on the observations that sequential data fetch is far more i T efficient than random access, that improving seek dis- tances produces only marginal performance improve- 0 ments, and that the increasingly powerful processors and 1991 1995 1999 2003 large memories in storage systems have ample capacity Year Disk was Introduced to reorganize the data layout and redirect the accesses so as to take advantage of rapid sequential data trans- Figure 1: Time Needed to Read an Entire Disk as a fer.
    [Show full text]