云平台网络节点单体监控信息分布式存储方法

Distributed Storage Method for Individual Monitoring Information of Cloud Platform Network Node

针对监控信息储存负载不均衡、丢包率高等问题,提出云平台网络节点单体监控信息分布式存储方法。将监控信息源当作一个由多个事件构成的集合,通过事件概率分布计算信息熵的最大值,根据该值完成监控信息压缩和去冗余处理;利用异地数据备份恢复算法计算远程服务器负载,结合负载情况实时调整备份任务分配量,确保负载均衡,避免信息丢失;建立包括软件服务、平台服务和基础设施服务的三层云储存架构,分析不同层次的功能;将分布式储存过程分为访问层、连接层与归档层,通过计算信息访问概率,判断监控信息是否满足储存要求,对于满足的信息直接储存,针对不满足条件的信息,利用粒度率调节弹性期望值,减少信息储存拥堵现象,实现分布式储存。实验表明,所提方法压缩效果好,能够确保储存节点的负载均衡,减少了监控信息丢失现象。

Aiming at the problems of unbalanced load and high packet loss rate of monitoring information storage,a distributed storage method for individual monitoring information of cloud platform network node is proposed.The monitoring information source is regarded as a set composed of multiple events,the maximum value of information entropy is calculated through the event probability distribution,and the monitoring information compression and redundancy removal are completed according to the value.The remote server load is calculated by using the remote data backup and recovery algorithm,and the backup task allocation is adjusted in real time according to the load situation to ensure load balance and avoid information loss.We establish a three-tier cloud storage architecture including software services,platform services and infrastructure services,and analyze the functions at different levels.The distributed storage process is divided into access layer,connection layer and archive layer.By calculating the information access probability,it can judge whether the monitoring information meets the storage requirements.For the information that meets the requirements,it can be stored directly.For the information that does not meet the conditions,it can use the granularity rate to adjust the elastic expectation value,reduce the information storage congestion,and realize distributed storage.Experiments show that the proposed method has good compression effect,can ensure the load balance of storage nodes,and reduce the loss of monitoring information.