| The complexity of developing multiplayer games, along with their popularity, has grown tremendously in the recent years. The most complex of these, Massively Multiplayer Games (MMOGs), require developers to deal with many issues, such as scalability, reliability and cheat prevention. Although individual solutions to these problems exists, very little academic work has been done to address all these issues simultaneously. In addition, experimentation in these areas can require a significant implementation effort.;The efficiency of this framework is illustrated through the use of Mammoth, a massively multiplayer research framework. Using experimental data from human players, artificial players (NPC) were built and used to stress test and gather performance data. Analysis of this data demonstrated that load balancing provides important scalability benefits while very little overhead is incurred from the fault tolerance and cheat prevention systems.;In this work, we present Journey, a unified framework that address all these issues in a simple, modular and efficient architecture leveraging replicated objects. Scalability is addressed through the use of a dynamic cell load-balancing strategy while fault tolerance and cheat prevention are achieved by leveraging existing replicated objects in the system. The proposed framework is implemented using numerous enhancements not found in traditional replication, like obstacle-aware partitioning and remote procedure call systems. |
