The OMG Data Distribution Service (DDS) is an open, real-time publish/subscribe middleware standard, which has been widely adopted in many latency-sensitive and mission-critical industrial IoT applications. However, deploying and managing large-scale distributed DDS applications is tedious and laborious. As a successful container orchestration platform for distributed applications in the cloud, Kubernetes (k8s) is a promising solution for DDS-based systems. However, the feasibility of running DDS applications in a k8s environment, and the overhead of different k8s virtualization network architectures on DDS application performance has not been systematically studied. To address this, we designed an automated benchmark framework, analyzed potential bottlenecks of several popular k8s virtual network plugins, and conducted a comprehensive set of experiments to demonstrate DDS performance variation under different QoS and k8s network configurations.