Microservice Indicator Prediction Method Based on STE and CNN-BiLSTM
Yunhao Yang, Ying Jiang · 2023
Due to the extensibility and continuous evolution of microservice architecture, there are a lot of uncertainties in the microservice system, which brings great risks to the reliability of the service. Indicator prediction plays an important role in service reliability. If the predicted value exceeds the safe range, alarms are generated and measures are taken to prevent faults. Therefore, a microservice indicator prediction method based on SET and CNN-BiLSTM is proposed. Symbolic transfer entropy (STE) is used to analyze the nonlinear causality, and a prediction model based on CNN-BiLSTM is established. The simulation results show that this method can capture the causal relationship between the indicators with nonlinear relationship effectively and improve the prediction accuracy.