In order to evaluate our optimized algorithm in practice, we present the experimental
setup for evaluating. We set up a cluster of 16 Server PC machines, each of them has a
Dual Core AMD Opteron 2.00 GHz CPU, 73GB hard drive, Inter 82551 10/100 Mbps
Ethernet Controller, 2GB memory and Ubuntu10.10 server OS. All nodes are connected to a 100 Mbps Ethernet switch.