Enforcing replica determinism

Abstract

The necessity to enforce replica determinism in non-trivial systems has been shown in the previous chapter. This chapter is concerned with methodologies and implementations that are appropriate to enforce replica determinism in the context of fault-tolerant real-time systems. Since it is impossible to achieve ideal replica determinism, all implementations—as presented in this chapter—relax the semantics to achievable cases of replica determinism. Achievable semantics are either based on epsilon common knowledge or on concurrent common knowledge.