An Attribute Similarity-Based Approach to Compress and Merge Alarms of Multiple Microservice Applications

In microservice architecture, due to the fine-grained service splitting, a large number of alarm messages will be generated, which, if not processed, will lead to the operation and maintenance personnel drowning in the sea of alarms, thus missing the real critical problems. Therefore, this paper proposes an attribute similarity-based alarm compression and merging method for multiple micro service applications, which aims to reduce redundant alarms and improve the accuracy and processing efficiency of alarms. In this study, the seasonal and trend characteristics of monitoring indicators are eliminated by the time-series decomposition algorithm STL, and the multivariate time-series prediction of the residual series of multiple indicators is performed based on the vector autoregressive model VAR, and whether the microservices are abnormal is judged based on the Euclidean distances of the vectors; according to the characteristics of the ASON alarms of an actual electric power communication network, the communication alarm correlation model based on the hierarchical attribute similarity is established. Finally, a compression and merging method of multiple microservice application alarms based on attribute similarity is used in a real power system communication network. The test results show that by merging similar alarms, the number of alarms is reduced, which helps the operation and maintenance personnel to filter out the real problems that need attention from a large number of alarm messages. Moreover, the alarm compression and merging method based on attribute similarity has significant technical advantages in improving operation and maintenance efficiency, optimizing alarm management, and reducing operation and maintenance costs.