Consistency and instanced massively multiplayer games

点击翻译按钮获取中文摘要

Multiplayer games allow many human players to experience a game collectively, and have become an extremely popular genre. Ideally, these worlds are large, allowing a large number of people from across the world to interact. Network and computational resources, however, do not scale to larger (massive) sizes easily for distributed game environments, and current solutions involve various restrictions on virtual game world size, player interactivity, or ability to ensure each player simultaneously sees the same state. We base our proposal on an extension of current "instanced" game models, where players do not all play concurrently, rather a smaller group of at most around 100 players participate in a game, with each game instance quite separate from others. Instances can be accessed from a main game world or virtual "lobby" environment that allows network connected players to find other willing game participants. By extending the available interactions between disparate instances, and between instances and the game world or lobby we explore new game functionality as well as establish further techniques for extending game scalability, allowing more players to simultaneously play a game while ensuring an efficient underlying implementation. Our work will provide suitable new game consistency models, as well as experimental implementation in our Mammoth research game environment, and also a commercial game lobby environment.