RSX: Reproduction Scenario Extraction Technique for Business Application Workloads in DBMS

Reproducing software defects plays a crucial role in the process of diagnosing and resolving bugs. While reproducing bugs, programmers may apply additional analysis tools to locate the point at which the program has gone astray. In a database management system for handling massive concurrent workloads from business application, developing a reproduction scenario may be challenging due to complex transactional dependency among workload executions. In this paper, for SAP HANA Quality assurance, we introduce a technique to extract a lightweight reproduction scenario which contains the relevant sequence of database operations for retriggering a target system failure. Our approach comprises two phases: (1) transforming a historical transactional information into a compact data structure, (2) extracting a reproduction scenario to isolate the symptom. We evaluate our technique against real functional bugs detected during the SAP HANA release validation and highlight its effectiveness for handling complex symptom analysis.