Synchronous replication seems ideal for high availability. However, practical constraints make it very difficult to do. For example, a network connection requires extremely strong (especially during peak hours). Reduced network bandwidth or resource performance beyond terrible. Moreover, if all writes replicated copies of databases or remote networks and binding off, I want to write, has no effect on application availability. Of course, you can use the small local, while mandatory are written copies of the server type is optional. But mandatory and optional settings can be better served regular master-master configuration, which provides partial failure of the functionality of the deferred queue.
Synchronous replication is most important sites of the network capacity, where the distances between geographical locations (involved in replication)
This means that the application of the synchronous replication sites, which are located at a long distance is not easy. You can do a lot of things, such as bandwidth, logical corruption, some of the websites listed below, a long series of data errors and performance. These are common problems in the application. If you apply the same configuration is consisting of a network of private places where you can get the benefit of all, and solve simple problems.