Taint analysis is widely used for tracking sensitive data. However, state-of-the-art taint analyzers face challenges in recall, scalability, and precision when applied to industrial microservices. To overcome these challenges, we present a field-based static taint analysis approach, which does not distinguish different instances of the same type but distinguishes fields of the same kind for tracing sensitive data on industrial microservices. The experimental results demonstrate that our approach is practical in industrial scenarios.