Interference Detection of Colocation Workloads in Cloud Sidecar Clusters

In modern microservice architecture, the colo cation of online and offline services is becoming increasingly prevalent. While the Istio service mesh, with its sidecar proxy model, facilitates the orchestration, monitoring, and security of microser-vices. However, the colocation of diverse workloads can lead to interference issue, affecting the performance and reliability of the system. This paper presents a novel interference detection mechanism tailored for colocation deployments on mesh sidecar clusters. Our approach leverages the robust observability and traffic control features provided by service mesh to monitor and analyze the behavior of services in real-time, identifying interference scenarios and their impact on service performance. We have carried out the colocation interference experiment, and the relevant metrics are collected and analyzed. In addition, we proposed an interference detection algorithm based on HBOS algorithm to classify experimental data. Experimental results show that our detection algorithm has a high accuracy.