基于链路聚合的黑盒网络故障模型

Black Box Network Fault Model Based on Link Aggregation

针对互联网网络故障预判能力差、预判精度低的问题,在OPA模型基础上,将网络输入、输出端口聚合,并将网络黑盒化,建立数学模型。在已建立的模型的基础上,引入网络约束条件,添加极小化代价函数及负载分配控制策略,分析黑盒网络整体负载情况。通过分析链路负载与节点流量之间数量关系和相邻天数内网络负载比的概率分布参数情况,得出网络输入与输出负载比,通过所得参数判断网络故障发生的概率。通过网络模型抽象化和黑盒化,屏蔽网络内部复杂性,简化网络故障分析,从而整体判断网络故障发生的概率。通过实验仿真,验证了网络模型的有效性及算法的可行性及优越性,一定程度上提高了互联网网络故障预判能力。

Aiming at the problem of poor fault prediction ability and low prediction accuracy of Internet network,based on OPA model,the network input and output ports were aggregated,and the network was black boxed by which a mathematical model was established. Based on the model,the overall load of the black box network was analyzed by adding the network constraint conditions and the minimization cost function as well as the load allocation control strategy. By analyzing the quantitative relationship between link load and node traffic and the probability distribution parameters of network load ratio in adjacent days,the network input and output load ratio was obtained,and the probability of network failure was judged by the parameters. Through the abstraction and black-boxing of the network model,the internal complexity of the network was shielded,and the network fault analysis was simplified,so as to determine the overall probability of network fault. Through experimental simulation,the effectiveness of the network model and the feasibility and superiority of the algorithm are verified,and the ability of Internet network fault prediction is improved to some extent.