分布式系统数据一致性和并发性优化研究

Research on Data Consistency and Concurrency Optimization of Distributed System

为满足云存储的高数据读写吞吐量需求,提出一种用于分布式系统的数据锁和快速定位数据锁方法。该锁结构通过细化数据覆盖的颗粒度,在确保数据一致性的前提下允许数据共享。以吞吐量和冲突率为指标进行仿真实验,结果证明,该数据锁定位方法可以向上层提供高性能的随机数据存取服务。

A novel lock scheme applied in distributed systems is proposed in order to fulfill the requirement of huge data throughput, The lock structure refines the granularity of data coverage and allows data sharing in the premise of data consistency while the lock localization algorithm decreases the time to look up lock objects. These methodologies are evaluated in terms of throughput benchmark and lock collision ratio, and simulation result demonstrates that with these optimizations the storage layer can provide high-performance random access to data retrieval requests from upper layer.