Maintaining Cache Consistency in RDMA-based Distributed Key-Value storage System
Zhuohan Hou, Qianyi Zheng, Yuanfeng Song, Xiaowen Nie · 2024
With the development of big data and artificial intelligence, the proliferation of data brings new challenges to the key-value storage systems. The Remote Direct Memory Access (RDMA) can bring clear performance benefits to the key-value storage systems. This research addresses the optimization cache consistency of ScaleStore. We propose a novel method, Cache consistency algorithm, R-scale(Reoptimization ScaleStore). It adopt a new paradigm, which is the separation of storage and compute, develop a novel write-back buffer approach and the asynchronous remote invocation. Compared to ScaleStore's cache consistency, the R-scale’s write performance is improved by 20%.