Your motivation to design and implement a solution for a failure to commit depends on the probability of a failure occurring in your particular application. If the likelihood of a write-write conflict is very very small, then it may be acceptable to simply display an error message to the user, abort the transaction, and force the user to begin the business transaction again. On the other hand, when losing the user's changes is an unacceptable solution, especially when it may occur relatively frequently, you will want to provide a solution to recover from a failure to commit.