摘要
本文介绍了一种full-2码的虚拟顶点简单图表示法,简化了双容错数据布局判定定理,最优冗余数据布局定理和双容错数据布局的构造.本文还提出了一种基于完全二部图(对应二维奇偶校验码)的完全1-因子分解的双容错数据布局构造方法,可构造高扩展性双容错数据布局BG-HEDP.与B-CODE等同类双容错数据布局相比,BG-HEDP同样具有更新代价最优、高可靠性和低编码/解码复杂度的优点,冗余率接近最优,而扩展性更好.
We present a ＂virtual node＂ simple graph representation for full-2 code （corresponds to complete graph）, this representation simplifies the double-erasure-correcting data layout judgment theorem, the optimal redundancy data layout theorem and the construction of B-CODE.We also present a data layout construction method based on PIF of complete bipartite graph （corresponds to 2d parity code）, this method can produce highly extensible double-erasure-correcting data layouts （BG-HEDP）. Compared with other data layouts, such as B-CODE, BG-HEDP also has optimal update penalty, high reliability and low encoding/decoding complexity,its redundancy is very close to optimal value, while it is superior in extensibility to others.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2006年第B12期2447-2450,共4页
Acta Electronica Sinica
基金
国家自然科学基金(No.906120001)
天津市科技发展计划重点项目(No.043800311
No.04315111-14)
关键词
磁盘阵列
双容错编码
数据布局
完全二部图
完全1-因子分解
RAID（Redundant Array of Independent Disks）
2-erasure-codes
data layout
complete bipartite graph
perfect 1-factorization
