GBLT: Load Distribution Method for High Density by User Behavior in MMO Virtual Environment

This paper focuses on MMO(massively multiplayer online), a large-scale virtual environment, and we propose GBLT(group based load-distribution technique)-a load distribution method with P2P(peer-to-peer)-that can maintain scalability under the high density of user in virtual environment. For a high scalability on a rapid population change, a lot of systems using P2P is researched. However, the problem of some MMO is that userpsilas group behavior generates high density situation frequently. Many users may form group, and the existing methods manage a virtual environment by dividing geographically between peers. In MMO, a user high density may be generated because the users who participate to a same group are mutually adjacent on a virtual environment. Then, Many messages concentrate on a part of peer if the user high density situation is generated. The existing methods cannot maintain the performance in the high density situation because such a situation is not considered. Then, we propose a load distribution method GBLT considering an incrementation of messages by such a user high density on virtual environment. Then, new manager who manages group is added to the system when the group is formed in GBLT. As a result, an existing managerpsilas load is decreased. Therefore, GBLT can be adjusted to the high density. We confirm that GBLT can maintain high scalability under the local high density situation. Simulation shows that the average number of user managed by server is suppressed to low.