摘要
数据去重的过程中,云存储系统会产生大量的计算机调度冲突。针对去重操作调度冲突问题,提出一种基于长短期记忆网络(LSTM)机器学习方法的预测模型,可根据历史操作预测服务器负载情况,由此给出操作序列建议,对服务器进程进行合理调度并实施去重操作。并与基于随机模拟仿真数据的操作调度进行了对比,实验结果表明,本方案在提高服务器去重操作执行效率方面具有优势,能够降低服务器的性能开销。
In the process of data deduplication,a large number of computer scheduling conflicts will be produced by cloud storage system.In order to solve the deduplication scheduling conflict problem,a predictive model based on Long Short-Term Memory(LSTM)machine learning method is proposed.It can predict server load based on historical operations.The operation sequence is suggested,and perform deduplication according to the sequence.Compared with operation scheduling based on stochastic simulation data,the experimental results show that the scheme has advantages in improving the efficiency of server deduplication operation.It can reduces the performance overhead of the server.
作者
穆雪莲
咸鹤群
MU Xue-lian;XIAN He-qun(College of Computer Science and Technology,Qingdao University,Qingdao 266071,China)
出处
《青岛大学学报（自然科学版）》
CAS
2021年第1期25-28,共4页
Journal of Qingdao University(Natural Science Edition)
基金
山东省自然科学基金(批准号:ZR2019MF058)资助。
关键词
云存储
数据去重
调度优化
LSTM
预测模型
cloud storage
deduplication
schedule optimization
LSTM
prediction model
