Database systems have used a client-server computing model to support shared data in distributed systems such as Web systems. To reduce server bottlenecks, each client may have its own cache for later reuse. This paper suggests an efficient cache consistency protocol based on a optimistic approach. The main characteristic of our scheme is that some transactions that read stale data items can not be aborted, because it adopts a re-ordering mechanism to enhance the performance. This paper presents a simulation-based analysis on the performance of our scheme with other well-known protocols. The analysis was executed under the Zipf workload which represents the popularity distribution on the Web. The simulation experiments show that our scheme performs as well as or better than other schemes with low overhead.