{"title":"CPM:基于对象存储集群的协同电源管理","authors":"Q. Wei, B. Veeravalli, Zhixiang Li","doi":"10.1109/MASCOT.2009.5366151","DOIUrl":null,"url":null,"abstract":"Disk idle periods in server workload are short, which significantly limits the effectiveness of underline disk power management. To release this limitation, we present a Cooperative Power Management (referred to as CPM) scheme to save energy with performance guarantee for object-based storage cluster. CPM reclaims idle memories of neighboring Object-based Storage Devices (OSDs) over high speed network as remote cache to store evicted objects. Then requests missed in local cache could be hit by remote cache, and local disk does not necessarily spin back up to service these requests. Hence, CPM can artificially create long idle periods to provide more opportunities for underlying disk power management. CPM minimizes the risk of performance and energy penalty by spinning down disks only when predicted idle period is long enough to justify statetransition energy. Our rigorous experiment results conclusively demonstrate that CPM can dynamically adapt to workload changes and outperform existing solutions in terms of energy saving and performance for large-scale OSD cluster.","PeriodicalId":275737,"journal":{"name":"2009 IEEE International Symposium on Modeling, Analysis & Simulation of Computer and Telecommunication Systems","volume":"34 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2009-12-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"CPM: Cooperative power management for object-based storage cluster\",\"authors\":\"Q. Wei, B. Veeravalli, Zhixiang Li\",\"doi\":\"10.1109/MASCOT.2009.5366151\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Disk idle periods in server workload are short, which significantly limits the effectiveness of underline disk power management. To release this limitation, we present a Cooperative Power Management (referred to as CPM) scheme to save energy with performance guarantee for object-based storage cluster. CPM reclaims idle memories of neighboring Object-based Storage Devices (OSDs) over high speed network as remote cache to store evicted objects. Then requests missed in local cache could be hit by remote cache, and local disk does not necessarily spin back up to service these requests. Hence, CPM can artificially create long idle periods to provide more opportunities for underlying disk power management. CPM minimizes the risk of performance and energy penalty by spinning down disks only when predicted idle period is long enough to justify statetransition energy. Our rigorous experiment results conclusively demonstrate that CPM can dynamically adapt to workload changes and outperform existing solutions in terms of energy saving and performance for large-scale OSD cluster.\",\"PeriodicalId\":275737,\"journal\":{\"name\":\"2009 IEEE International Symposium on Modeling, Analysis & Simulation of Computer and Telecommunication Systems\",\"volume\":\"34 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2009-12-28\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2009 IEEE International Symposium on Modeling, Analysis & Simulation of Computer and Telecommunication Systems\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/MASCOT.2009.5366151\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2009 IEEE International Symposium on Modeling, Analysis & Simulation of Computer and Telecommunication Systems","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/MASCOT.2009.5366151","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0
摘要
服务器工作负载中的磁盘空闲时间很短,这极大地限制了底层磁盘电源管理的有效性。为了克服这一限制,我们提出了一种基于对象存储集群的协同电源管理(Cooperative Power Management,简称CPM)方案,在保证性能的前提下实现节能。CPM通过高速网络回收相邻osd (Object-based Storage device)的空闲内存作为远程缓存来存储被驱逐的对象。然后,在本地缓存中丢失的请求可能会被远程缓存击中,并且本地磁盘不一定会重新启动以服务这些请求。因此,CPM可以人为地创建较长的空闲时间,为底层磁盘电源管理提供更多机会。CPM只有在预测的空闲时间长到足以证明状态转换能量的合理性时,才会关闭磁盘,从而将性能和能量损失的风险降至最低。我们严谨的实验结果最终证明,CPM可以动态适应工作负载的变化,并且在节能和性能方面优于现有的大规模OSD集群解决方案。
CPM: Cooperative power management for object-based storage cluster
Disk idle periods in server workload are short, which significantly limits the effectiveness of underline disk power management. To release this limitation, we present a Cooperative Power Management (referred to as CPM) scheme to save energy with performance guarantee for object-based storage cluster. CPM reclaims idle memories of neighboring Object-based Storage Devices (OSDs) over high speed network as remote cache to store evicted objects. Then requests missed in local cache could be hit by remote cache, and local disk does not necessarily spin back up to service these requests. Hence, CPM can artificially create long idle periods to provide more opportunities for underlying disk power management. CPM minimizes the risk of performance and energy penalty by spinning down disks only when predicted idle period is long enough to justify statetransition energy. Our rigorous experiment results conclusively demonstrate that CPM can dynamically adapt to workload changes and outperform existing solutions in terms of energy saving and performance for large-scale OSD cluster.