GSM: a game scalability model for multiplayer real-time games

Current commercial real-time computer games increasingly provide game designs suitable for a high number of players, forming the class of massive multiplayer games (MMG). To support MMG, the scalability of network topologies becomes critically important, i.e., their ability to maintain the game service for an increasing number of players. This paper presents GSM - an analytical scalability model for a detailed investigation of massively multiplayer capabilities of different networking topologies. We use the GSM to discuss the suitability of the client-server and peer-to-peer topology, as well as our own concept of a multi-server topology, for the important classes of first person shooter (FPS) and real-time strategy (RTS) games whose designs are still rarely adopted in MMGs. We verify our analytical model in scalability experiments in which we were able to forecast the maximum numbers of players in a non-congested game session with un error of less than 7 %.