This paper investigates the performance improvement opportunities through a resiliency mechanism in the distributed primary/backup database system. Recognizing that a distributed transaction executes at several servers during its lifetime, we propose a resiliency mechanism which allows continuous transaction processing in distributed database server systems in the presence of a server failure. In order to perform continuous transaction processing despite failures, every state change of a transaction processing can be saved in the backup server. Obviously, this pessimistic synchronization may give more burdens than benefits to the system. Thus, the tracking needs not be done synchronously with the transaction progress. Instead, the state of all transaction processing in a system is saved periodically. This activity is referred to as a savepoint. Then, the question is how often the savepoint has to be done. We derive the optimal savepoint to identify the optimization parameters for the resilient transaction processing system.