-
1.
公开(公告)号:US5566348A
公开(公告)日:1996-10-15
申请号:US965113
申请日:1995-10-22
CPC分类号: G06F3/0601 , G06F12/127 , G11B27/002 , G06F12/123 , G06F2003/0697 , G11B2220/20 , G11B2220/41 , G11B2220/90
摘要: A method for managing Data Storage Medium (DSM) mount and demount decisions in an automated data storage library that dynamically optimizes both sequential and random data access workloads. The demount decisions adapt to time-varying characteristics in the relative workload mix. When the workload is primarily sequential, the mount and demount decision procedure favors longer mount residency for sequential access streams, reducing the robotic picker activity and reducing response time for mount requests. When the workload is predominantly random access, sequentially accessed DSM residency time is generally reduced and preemptive demounts are more readily implemented. The disclosed method provides for preemptive demounts and uses a two-element decision process to select either a Least Recently Used (LRU) or a Least Recently Mounted (LRM) decision parameter. The relative weights of the LRU and LRM decision tests are varied responsive to measured changes in data access workload characteristics.
摘要翻译: 一种在自动数据存储库中管理数据存储介质(DSM)安装和拆卸决策的方法,可动态优化顺序和随机数据访问工作负载。 拆卸决策适应相对工作负荷组合中的时变特征。 当工作负载主要是顺序时,安装和拆卸决策过程有利于顺序访问流的更长的安装驻留,减少了机器人选择器活动并减少了安装请求的响应时间。 当工作负载主要是随机访问时,通常减少顺序访问的DSM驻留时间,并且更容易地实现抢占式卸载。 所公开的方法提供抢占式拆卸并且使用双元件决策过程来选择最近最少使用(LRU)或最近最少安装(LRM)最小安装(LRM)决策参数。 响应于测量的数据访问工作负载特征的变化,LRU和LRM决策测试的相对权重变化。