Large-scale reactive systems developed by multiple organizations are particularly vulnerable to late integration, i.e., Big-bang, where hidden architectural and operational mismatches had only been revealed at the final system assembly. The paper reports on an industry experience of a one-year multi-partner project that failed due to delayed integration, limited observability, and insufficient environment parity. From an architecture-operation perspective, we analyze the causes, highlighting how the system behaviors are undermined. Based on the analysis, we describe Integractor, an actor-centric, messagedriven early-integration approach that combines standardized control APIs, mock-first development, and broker-based communication to enable continuous system integration across distributed teams. Furthermore, we evaluated the approach using scenario-based architectural analysis and a simulated integration environment that reflects realistic operational constraints. Importantly, from this failure we derive a set of practical lessons learned to embed xOps concerns directly into architectural design decisions.
- Learning from a Big-Bang Integration Failure in a Distributed Reactive System: An xOps Industry Case Study
- Alessandro Aneggi - Free University of Bozen-BolzanoXiaozhou Li - Free University of Bozen-BolzanoAndrea Rosani - Free University of Bozen-BolzanoAndrea Alexander Janes - Free University of Bozen-BolzanoOswald Lanz - Free University of Bozen-Bolzano
- Proceedings: 2026 IEEE 23rd International Conference on Software Architecture Companion, ICSA-C 2026 , pp.576-583
- 9798331557409
- 9798331577872
- 2768-427X
- 2768-4288
- 23rd IEEE International Conference on Software Architecture, ICSA 2026 (Amsterdam, 22/06/2026–26/06/2026)
- International Conference on Software Architecture Companion
- IEEE
- 8
- 9798331557409
(UNIBZ)99706882
991007420494901241 - 2-s2.0-105048466694
- Faculty of Engineering
- English
- Conference proceeding
- Aneggi A, Li X, Rosani A, Janes A, Lanz O