Work

Research on Microservice Anomaly Detection Technology Based on Conditional Random Field

Year: 2019
Type: conference-paper
Abstract: Recently, the existing microservice anomaly detection methods which mainly monitor the system metrics and analyze the system logs have low efficiency and consume much time. This paper proposes a microservice anomaly detection method based on Conditional Random Field(CRF). The method takes the collected system parameters such as CPU utilization, memory utilization, bandwidth occupancy as the characteristic values of the observation sequence, and labels the abnormal types of the sequences corresponding to the feature values, and then creates the microservice fault matrix. Operators can quickly retrieve the specific anomaly information of the microservice error according to the microservice fault matrix. Compared with the traditional Hidden Markov Model-based detection method, this method can find the abnormal sequence corresponding to the observation sequence in the global scope. The experimental results show that the method proposed by this paper can accurately find the faults in the microservice system, and the accuracy and recall rate are relatively high.
Language: English
Open Access status: diamond
OpenAlex ID: W2949078428
FWCI: 0.342
Cites: 2
Cited by: 6
Related to: 10