基于XOR的TAR-CAU数据更新方法

TAR-CAU: An XOR-Based Data Update Scheme

  • 摘要: 在基于纠删码的云存储系统中,数据更新的性能往往受到网络带宽的限制。鉴于此,提出了基于异或(XOR)的TAR-CAU数据更新算法,该算法基于以下设计原则:1)利用数据更新量普遍较小的特点,将多条带的数据更新打包处理,减少网络往返次数,加快数据传输效率;2)采用基于XOR的更新,提高编解码效率。仿真实验和本地集群实验结果表明,相比于CAU算法,在数据更新量较小时,TAR-CAU算法能够提高至少44%的数据更新吞吐量。

     

    Abstract: In an erasure-coded cloud storage system, the performance of data updates is often limited by network overhead. To end this, based on Cross-Rack-Aware Update (CAU) data update scheme, we propose an XOR based Tape ARchive Cross-Rack-Aware Updates (TAR-CAU) data update scheme. TAR-CAU is designed in terms of the two design primitives: 1) as the updates are small, we can pack several data blocks of data update into one block to reduce the number of network round trips, and accelerate the data transmission; 2) XOR-based data update scheme is used to accelerate data encoding and decoding. The simulation experiments and local cluster experiments show that, when the data updates are small, TAR-CAU can increase the data update throughput by at least 44% compared with the CAU.

     

/

返回文章
返回