Summary

The article discusses the challenges of testing microservices, particularly those influenced by complex inputs like payment cost estimation. It emphasizes the importance of replay testing using Apache Spark to validate service behavior across historical data, enabling teams to assess the impact of code changes before deployment. The author outlines how to separate core logic from transport mechanisms to facilitate this testing approach. The article also highlights the use of regression testing and what-if analysis to model future rule changes effectively.

💭 Your Thoughts

Using Spark job for regression test is a new use case for me. We usually run Spark job to produce production data. I think the only downside is how to make sure the logic in microservice and in Spark are the same, also the input data versions etc.

Data Problems Addressed

Technologies Referenced