Fault-tolerance mechanisms are particularly difficult to integrate into complex distributed systems. This paper builds on the principles we had developed in our DSN'03 paper and proposes a generic reflective mechanism to seamlessly support replication in complex off-the-shelf systems (OS, middleware, libraries). Key to the paper is a case study that looks at a commercial CORBA implementation and demonstrates the benefits of the approach on an industry-grade platform. It was presented at DSN, the main conference for dependability computing. (Acceptance rate in 2005: DCCS track 54/204 ~ 26.8%). RAE_import_type : Conference contribution RAE_uoa_type : Computer Science and Informatics