Communication‐induced checkpointing (CIC)
can prevent a potential
domino effect
resulting from cascading rollbacks in a distributed computation. Moreover, it can be enhanced to achieve a stronger property, called
rollback‐dependency trackability (RDT)
, which allows all hidden rollback dependencies to be online trackable. In this article, we survey some important CIC and RDT protocols by introducing their basic concepts as well as by discussing their practical impacts.