Administrators can reduce latency, but at a performance cost. If each of the six servers replicated with each of the other five servers, the system could converge, or be universally in sync, about once an hour (assuming again that replication occurred every hour). Figure 3 shows such a fully enmeshed design. In this fully enmeshed design, each server must maintain repli- cation agreements with five other servers, and must replicate with each server every hour. This much replication, particularly in a busy database application, would likely slow response so much that the performance gain achieved by creating a server farm would be lost. Each office might require two servers just to maintain repli- cation and meet users’ needs. Although fairly easy to implement, the server farm technique has a point of diminishing returns.