Hai-Hao Tang, Hu Chen, Jiangfeng Peng · 2009
Multi-core computers have become increasingly popular, in order to improve the performance of index in database systems on multi-core computers, and this paper presents an alternative algorithm of linear hashing and multithreaded disk buffer on multi-core computers. In this algorithm, records can be inserted into hash table concurrently by different threads. The key of the parallel insertion is to make traditional insertion into batch insertion. Multithreaded disk buffer is developed for multithreaded linear hashing, making multithreads allocate and release memory simultaneously as much as possible. It is applied to maximize memory utilization and to minimize the number of disk accesses. Our experimental result shows that, compared to traditional serial linear hashing, multithread linear hashing achieves a higher performance on multi-core computers.
Cited by 1
Mahesh Kallahalla, Peter J. Varman · 1998
We present an on-line algorithm for prefetching and caching for a multiple-disk parallel I/O system. We introduce the notion of write-back whereby blocks are relocated between disks during the course of the computation. Write-back allows the layout to be altered dynamically to suit access patterns in different parts of the reference string. The algorithm is analyzed using the standard parallel disk model with D parallel disks and a shared I/O buffer of size M blocks. We show that any on-line algorithm with M-block lookahead using deterministic write-back and buffer management policies must have a competitive ratio of \\Omega\\Gamma D). We therefore present a randomized algorithm, RAND-WB, that uses randomized write-back and attains a competitive ratio of \\Theta( p D), the best achievable by any on-line algorithm with M-block lookahead. Simulations show that RAND-WB obtains considerable performance benefit even when the data has just a small amount of re-use.
Cited by 0
2008
We provide a competitive analysis framework for online prefetching and buffer management algorithms in parallel I/O systems, using a read-once model of block references. This has widespread applicability to key I/O-bound applications such as external merging and concurrent playback of multiple video streams. Two realistic lookahead models, global lookahead and local lookahead, are defined. Algorithms NOM and GREED based on these two forms of lookahead are analyzed for shared buffer and distributed buffer configurations, both of which occur frequently in existing systems. An important aspect of our work is that we show how to implement both the models of lookahead in practice using the simple techniques of forecasting and flushing. Given a D-disk parallel I/O system and a globally shared I/O buffer that can hold upto M disk blocks, we derive a lower bound
Cited by 0
Robert F. Hodson · NASA Technical Reports Server (NASA) · 1992
The goal of this project is to develop an Application Specific Integrated Circuit (ASIC) for use in the control electronics of the Spacecraft Optical Disk Recorder (SODR). Specifically, this project is to design an extendable memory buffer controller ASIC for rate matching between a system Input/Output port and the SODR's device interface. The aforementioned goal can be partitioned into the following sub-goals: (1) completion of ASIC design and simulation (on-going via ASEE fellowship); (2) ASIC Fabrication (at ASIC manufacturer); and (3) ASIC Testing (NASA/LaRC, Christopher Newport University).
Cited by 0Open access
Feng Wang, Scott Brandt · 2001
Disk is approaching its performance limits in next decade. The performance gap between the main memory and the secondary storage device tends to be large. Research for new storage device is underway, among which the probe-based(MEMS) storage device [1, 2] is promising. It provides much better response time and bandwidth comparing to the modern disk. In this paper, we exploit the performance improvement using MEMS device as the disk write buffer. Our result shows that several hundred MBs MEMS device will eliminate almost all the raw write traffic to disk. The average response time will be improved by factor of 8 to 10 comparing to modern disk.
Cited by 0
Raju Rangaswami, Ž. Dimitrijević, Edward Yi Chang et al. · 2004
The performance of streaming media servers has been limited due to the dual requirements of high throughput and low memory use. Although disk throughput has been enjoying a 40% annual increase, slower improvements in disk access times necessitate the use of large DRAM buffers to improve the overall streaming throughput. MEMS-based storage is an exciting new technology that promises to bridge the widening performance gap between DRAM and disk-drives in the memory hierarchy. We explore the impact of integrating these devices into the memory hierarchy on the class of streaming media applications. We evaluate the use of MEMS-based storage for buffering and caching streaming data. We also show how a bank of k MEMS devices can be managed in either configuration and that they can provide a k-fold improvement in both throughput and access latency. An extensive analytical study shows that using MEMS storage can reduce the buffering cost and improve the throughput of streaming servers significantly.
Cited by 22
Xiongzi Ge, Dan Feng, David Hung-Chang Du · 2011
Much research has been conducted on energy efficient cache buffer management for disk based storage systems. Some of them use greedy prefetching technique to artificially increase disk idle intervals if there are a large number of known future requests. However, this might result in sub-optimal solution by not exploiting the relationship between I/O access pattern (sequential/random) and application pattern (CPU required for computing time). In a CPU-bound application, by explicitly taking into account this relationship it may reduce energy conservation by up to 50% and increase power cycle number by 100% compared to an existing efficient prefetching scheme without this consideration. In this paper, we consider the tradeoff between disk power consumption, performance guarantee and disk reliability all together by proposing a Disk characteristic based Power-Optimal Prefetching (DiscPOP) scheme. Specifically, we make two contributions: (i) A theoretical model is conducted to analyze energy-efficient cache buffer management in disk I/O system and it is formulated as an optimization problem. We have shown it can be solved via an Integer Linear Programming (ILP) technique. (ii) We propose a simple Divide-and-Conquer based offline algorithm named Greedy Partition (GP) to divide the problem into several small ones and solve them separately via an ILP solver. We use trace-driven simulations to evaluate our proposed scheme. The results show GP outperforms the traditional aggressive prefetching by up to 29.2% more disk energy conservation and 20.6% power cycle reduction.
Cited by 3
Raymond T. Ng · Open Collections · 2009
In this thesis, we study the problem of how to maximize the throughput of a multimedia system, given a fixed amount of buffer space and disk bandwidth, both pre-determined at design-time. Our approach is to maximize the utilization of disk and buffers. We propose two methods. First, we analyze a scheme that allows multiple streams to share buffers. Our analysis and simulation results indicate that buffer sharing could lead to as much as a 50% reduction in total buffer requirements. Second, we develop three prefetching strategies: straight forward prefetching (SP), and two types of intelligent prefetchings (IP1 and 1P2). All these strategies try to read a certain amount of data into memory before a query is actually activated. We demonstrate that SP is not effective at all, but both IP1 and 1P2 can maximize the effective use of buffers and disk, leading to as much as a 40% improvement in system throughput. We also extend the two intelligent prefetching schemes to a multiple disk environment.
Cited by 42Open access
Betev, Latchezar · CERN Document Server (European Organization for Nuclear Research) · 2021
Cited by 0
Jeffrey Koppanyi · Research Explorer (The University of Manchester)
Cited by 0
Afonso, Joao · CERN Document Server (European Organization for Nuclear Research) · 2024
<!--HTML-->The original design of EOSCTA inherited its cache eviction mechanism — stagerrm — from CASTOR. Although similar, CTA's use-cases were not the same as CASTOR's, leading to several operational issues. This motivated the creation of a new command — evict — better adapted to modern disk buffer management in CTA.\n\nIn this talk, we will discuss the issues that we faced with stagerrm and how the new evict command helps to fix them. We will also cover the new features brought by evict and how they simplify EOSCTA operations.
Cited by 0
Yang, Jinhai · cIRcle (University of British Columbia) · 1994
In this thesis, we study the problem of how to maximize the throughput of a multimedia system, given a fixed amount of buffer space and disk bandwidth, both pre-determined at design-time. Our approach is to maximize the utilization of disk and buffers. We propose two methods. First, we analyze a scheme that allows multiple streams to share buffers. Our analysis and simulation results indicate that buffer sharing could lead to as much as a 50% reduction in total buffer requirements. Second, we develop three prefetching strategies: straight forward prefetching (SP), and two types of intelligent prefetchings (IP1 and 1P2). All these strategies try to read a certain amount of data into memory before a query is actually activated. We demonstrate that SP is not effective at all, but both IP1 and 1P2 can maximize the effective use of buffers and disk, leading to as much as a 40% improvement in system throughput. We also extend the two intelligent prefetching schemes to a multiple disk environment.
Cited by 1Open access
David L. Livingston · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1988
In order to develop good error-control coding strategies for a read/write optical disk it is necessary to know the error characteristics of the media. Since the technology is in its genesis, empirical information on the error traits is not available. It is thus necessary to base the design of coding strategies on projected behavior and similar, well-established technologies. We examine the generation of a suitable error control strategy for an optical disk buffer which is to be used for the temporary or archival storage of large quantities of data. The buffer consists of multiple, read/write, magneto-optic surfaces which are independently accessable. Information is stored in a spiral eight-bit wide track; thus an appropriate error-control code must be capable of correcting symbol and burst errors. The properties of Reed-Solomon codes, which are well known for their packet protection capabilites, are considered for such an application. Since there are multiple surfaces, we present the possibility of coding across the surfaces. With this type of coding an entire surface can be lost without destroying the integrity of the data. Intermittent operation of a read/write device such as a laser could be corrected, thus providing a graceful degradation of operation. The interaction of independent access and simultaneous read/write capabilites with the coding scheme is considered.
Cited by 0
Anna Hat · 1992
This paper studies sensitivity of asynchronous algorithms in disk buffer cache memory. The algorithms allow for writing the files into the buffer cache by the processes. The number of active processes in the system and the length of the queue to the disk buffer cache are considered in sensitivity study. This information is obtained dynamically during the execution of the algorithms. The time for a block to become available for update in the buffer cache is given. The number of blocks available for update in the buffer cache is derived. The performance of the algorithms is compared. It is shown that the proposed algorithms allow for better performance than an algorithm that does not use the information about the system load.
Cited by 0
Jamel Gafsi, Ernst Biersack · 1997
A video server allows to deliver multiple video streams to different clients. To provide the required amount of storage and bandwidth, a video server must contain a large number of disks. Since the retrieval rate of the disk and the consumption rate of the clients differ, the data retrieved from disk need to be temporarily stored in main memory, which is an important cost factor in a video server. Video data must be retrieved from disks in such a way that neither buffer starvation nor overflow occurs for all concurrent video streams. In this paper we calculate the required buffer for the GSS scheduling algorithm for multiple streams retrieved from multiple disk server nodes. We prove analytically that shared buffer management reduces, in comparison with the dedicated buffer management, the required buffer by up to 50%. 1
Cited by 0
Anna Hác · 1992
Studies the sensitivity of asynchronous algorithms in disk buffer cache memory. The algorithms allow for writing the files into the buffer cache by the processes. The number of active processes in the system and the length of the queue to the disk buffer cache are considered in sensitivity study. This information is obtained dynamically during the execution of the algorithms. The time for a block to become available for updating in the buffer cache is given. The number of blocks available for updating in the buffer cache is derived. The performance of the algorithms is compared. It is shown that the proposed algorithms allow for better performance than an algorithm that does not use information about the system load.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Cited by 0
Kam-kwan Cleo Li · 2001
Cited by 0
nasa · 2019
The operating instructions for the magnetic bearings of a high-performance optical disk buffer are provided. Among the topics that are discussed are the following: front panel layout, turn-on procedure, shut-down procedure, and latch-up protection. Additionally, comprehensive engineering drawings are presented for the design.
Cited by 0
Johnson, Bruce, Downer, James, Anastas, George, Jr. et al. · NASA Technical Reports Server (NASA)
The innovation investigated in this project was the application of magnetic bearing technology to the translator head of an optical-disk data storage device. Both the capability for space-based applications and improved performance are expected to result. The phase 1 effort produced: (1) detailed specifications for both the translator-head and rotary-spindel bearings; (2) candidate hardware configurations for both bearings with detail definition for the translator head; (3) required characteristics for the magnetic bearing control loops; (4) position sensor selection; and (5) definition of the required electronic functions. The principal objective of Phase 2 was the design, fabrication, assembly, and test of the magnetic bearing system for the translator head. The scope of work included: (1) mechanical design of each of the required components; (2) electrical design of the required circuitry; (3) fabrication of the component parts and bread-board electronics; (4) generation of a test plan; and (5) integration of the prototype unit and performance testing. The project has confirmed the applicability of magnetic bearing technology to suspension of the translator head of the optical disk device, and demonstrated the achievement of all performance objectives. The magnetic bearing control loops perform well, achieving 100 Hz nominal bandwidth with phase margins between 37 and 63 degrees. The worst-case position resolution is 0.02 micron in the displacement loops and 1 micron rad in the rotation loops, The system is very robust to shock disturbances, recovering smoothly even when collisions occur between the translator and frame. The unique start-up/shut-down circuit has proven very effective.
Cited by 0
NASA Technical Reports Server (NASA)
The operating instructions for the magnetic bearings of a high-performance optical disk buffer are provided. Among the topics that are discussed are the following: front panel layout, turn-on procedure, shut-down procedure, and latch-up protection. Additionally, comprehensive engineering drawings are presented for the design.
Cited by 0Open access