With the rapid growth in the number of edge devices, cloud-edge collaboration has become increasingly prevalent, positioning computing resources at the network’s edge. This strategy not only reduces request latency for edge users but also alleviates network congestion typical of cloud computing, marking it as a crucial technology for IoT (Internet of Things) development. Microservice architecture, known for its high degree of decoupling, has emerged as a popular framework for large-scale applications. While cloud computing platforms are commonly utilized to host microservice systems, the ability of cloud-edge collaboration platforms to meet users’ QoS (Quality of Service) requirements remains uncertain. In this paper, we explore the performance of microservice systems within a cloud-edge collaboration context. We construct a computing platform that supports cloud-edge collaboration to deploy microservice systems in three configurations: cloud, edge, and cloud-edge collaboration, subsequently comparing their performance across these modes. Additionally, we examine the responsiveness of various microservices to different computing resources, aiming for efficient resource allocation. Our observability study further investigates QoS in scenarios where failures occur on cloud or edge servers. Our findings suggest that under normal load conditions, microservices in the cloud-edge collaboration mode outperform those in standalone cloud or edge configurations, exhibiting significant differences in CPU and network bandwidth sensitivity. This study contributes to the understanding of microservice deployment and management in cloud-edge collaboration scenarios, offering insights into optimizing performance and ensuring reliability under varying conditions.