Switch fault early warning method and device

Abstract

The invention discloses a switch fault early warning method and device. The method comprises the following steps: continuously collecting operation data from a plurality of switches through a simple network management protocol and storing the operation data into a time sequence database; in response to determining that the plurality of switches all operate normally based on the time sequence database, periodically executing feature engineering processing on the operation data to obtain a feature vector; determining a difference between the feature vector and a prediction vector in the comparator in response to the presence of the prediction vector in an LSTM predictor; and outputting early warning information through a real-time alarm in response to the difference exceeding a predeterminedthreshold. According to the invention, visual undifferentiated early warning can be provided for different types of switches, the personnel ability requirement is reduced, and the early warning timeis advanced so as to prevent accidents.

Description

technical field [0001] The present invention relates to the field of switches, and more specifically, to a method and device for early warning of switch failures. Background technique [0002] The switch is an important part of the Internet system and is used to connect various nodes in the network. Once it fails, it will directly affect the communication between the networks. The network transmission rate will slow down, the network will be interrupted, or the transmitted data will be wrong. In view of this, fault monitoring and early warning for switches is an important part of network management. The switch itself provides some runtime indicator information, which can be monitored by the operation and maintenance personnel, such as processor usage, packet loss rate, throughput, etc. Once an abnormal situation is found, it can be dealt with in time. [0003] However, the index information provided by the switch itself in the prior art cannot directly indicate whether it i...

AI Technical Summary

Problems solved by technology

[0003] However, the index information provided by the switch itself in the prior art cannot directly indicate whether it is faulty or not. It needs experienced network engineers to judge, which is not intuitive; different switch brands and models have inconsistent data, and special experience and knowledge are required. If the switch is upgraded , the corresponding parameters will change; in addition, the monitoring at this level is already an afterthought, and it can only be processed after the fault and cannot be warned in advance, and the lag is too strong

[0004] There is no effective solution to the problems of difficult judgment, complex changes, and strong hysteresis of switch failures in the prior art

Images