Efficient in-place update with grouped and pipelined data transmission in erasure-coded storage systems

详细信息    查看全文

• 作者：Xiaoqiang Pei ^{xiaoqiangpei@nudt.edu.cn}Author Vitae ; Yijie Wang ; ^{wangyijie@nudt.edu.cn}Author Vitae ; Xingkong Ma ^{maxingkong@nudt.edu.cn}Author Vitae ; Fangliang Xu ^{xufl89@nudt.edu.cn}Author Vitae
• 关键词：In-place ; Update ; Grouping ; In-time ; Lazy-update
• 刊名：Future Generation Computer Systems
• 出版年：2017
• 出版时间：April 2017
• 年：2017
• 卷：69
• 期：Complete
• 页码：24-40
• 全文大小：1995 K
• 卷排序：69

文摘

We propose a three-layer framework to support both single and multiple updates. We propose a workload-aware group technique to dynamically adjust the group size. We propose a distributed pipeline technique to distribute the computation. We propose a hybrid update technique to be compatible with the node failure. We conduct extensive experiments to confirm the advantages of our approach.