Ohira and Sawatari [Phys. Rev. E 58, 193 (1998)] introduced a simple model for computer network and showed that phase transition takes place between sparse and congested phase when the packet creation rate increases. First, we show the differences between deterministic and random routing strategy in this O-S Model. Then we generalize this model by relaxing the network topology and changing the geometry of the clustered hosts. We have identified features of the phase transition which are influenced by these changes. And we have already known that the network shows maximum information transfer and fluctuation near critical point. So we try to figure out a simple way to let the system become more effective by shifting the onset of phase transition.