基于分条技术的磁盘阵列自适应数据分布优化

    Optimization of Automatic-adaptive Data Partitioning in Striped Disk Array

    • 摘要: 为提高磁盘阵列的响应时间和系统吞吐量,提出了一种自适应选择分条单元的方法.根据请求到达率将文件分为2类:访问频繁的文件通过满足实际所需吞吐量的方法选择分条单元;访问量不大的文件则采用自适应方法,即在相同硬件条件下,分条单元的选择只依赖于请求大小和请求到达率.并且,对分条单元的计算作了一定简化,只需要通过一个简单的公式,就可快速计算出较优的分条单元大小.通过理论分析和实验测试证明:该方法在时间和空间上都具有较好的性能.

       

      Abstract: To minimize the disk array response time for every file access request and improve the disk system I/O throughput,a method to choose the right striping unit was proposed.In this method,files could be divided into two groups in accordance with request arrival rate.For the files accessed frequently,the right striping unit was chosen by meeting the throughput requirement.But for the other files,a auto-adaptive method was presented.Under the same hardware conditions,the choice of striping unit depended on only two parameters: request size and request arrival rate.Moreover,the calculation of the striping unit was also been simplified,then only through a simple formula,the near-optimal striping unit size could be quickly and easily calculated.Theoretical analysis and experimental results indicate that the method has better performance both in time and in space.

       

    /

    返回文章
    返回