Locally repairable codes enables fast repair of node failure in a distributed storage system. The code symbols in a codeword are stored in different storage nodes, such that a disk failure can be recovered by accessing a small fraction of the storage nodes. The number of storage nodes that are contacted during the repair of a failed node is a parameter called locality. We consider locally repairable codes that can be locally recovered in the presence of multiple node failures. The punctured code obtained by removing the code symbols in the complement of a repair group is called a local code. We aim at designing a code such that all local codes have a prescribed minimum distance, so that any node failure can be repaired locally, provided that the total number of node failures is less than the tolerance parameter. We consider linear locally repairable codes defined over a finite field of size four. This alphabet has characteristic 2, and hence is amenable to practical implementation. We classify all quaternary locally repairable codes that attain the Singleton-type upper bound for minimum distance. For each combination of achievable code parameters, an explicit code construction is given.
翻译:本地可修理的代码可以快速修复分布式存储系统中的节点故障。 代码中的代码符号存储在不同的存储节点中, 这样就可以通过访问存储节点的一小部分来恢复磁盘故障。 在修复失败节点时联系到的存储节点数量是一个称为地点的参数。 我们认为, 本地可修理的代码可以在多个节点故障的情况下在当地找到。 通过删除一个修理组的代码补充中代码符号获得的破解代码被称为本地代码。 我们的目标是设计一个代码, 使所有本地代码都有规定的最低距离, 从而可以在当地修复任何节点故障, 只要节点的总数少于容忍参数。 我们认为, 可以在四号的限定字段中定义线性本地可修复的代码数量。 这个字母特征 2, 因而可以实际执行。 我们将所有端端可修复的本地代码分类为最小距离的单质型顶端。 对于每种组合的可实现代码参数, 都会给出明确的代码构造 。