Catch Failures, Not Just Bugs!! Where QA Meets Chaos Engineering

Since e-commerce services operate 24 hours a day, there is a high likelihood of failures occurring due to various causes. Because thorough testing by the QA team alone is not enough to guarantee complete stability, Oliveyoung decided to introduce chaos engineering, which experiments with predictable failures in advance.

Chaos engineering is a methodology that intentionally induces failures to find vulnerabilities in order to increase a system's Resilience. Experiment types include simulating application, host, resource, network, AZ, region, and human error (People Attack) scenarios.

Among these, Oliveyoung selected Application Level testing as its top priority. To prevent the null exception issue that had occurred in the past, the team intentionally injects null values into API response values to verify whether the service operates normally by only hiding the specific affected area rather than causing the entire service to go down.

For test efficiency and process standardization, Oliveyoung adopted a proxy configuration approach using mitmproxy. This allows them to intercept HTTP(S) traffic and manipulate response data, reducing the hassle of manual work and building a systematic experimental environment.