With this article we want to identify the main scalability issues for the development of Massive Multi-Player Online Games. There is no generic architecture to achieve scalability for every problem. We must understand the nature of the problem in order to reach system scalability. Massive Multi-Player Online Games (MMOG) are conceived with the objective of massive use by a potentially geographically dispersed population. In their design we are faced with scalability challenges which are specific to the interactive modalities and the socio-technical scenarios we intend to enable [Fitch 2001]. The emergence of the Internet made possible the development of interactive distributed systems that can be accessed by thousands of users in virtually any part of the world. The scalability issues introduced by such a massive use must be considered in the system design. By scalability we mean the system fit capacity according to his loading charge, for example, accommodates increasing interaction volume, without significant degradation of quality service. It is commonly known that scalability can’t be secured if we only pay attention to some system parts. To achieve scalability in any kind of distributed system we must design all the components to achieve this goal. For example, a system that has high scalability in the simulation and low communication scalability may result in a poor scalable system, globally. To see the scalability problems in a MMOG we must understand the system dynamics and structure and what’s bound for. Looking at the existent types of MMOG – massive multi-player online role playing games, virtual environments, massive multiplayer real time strategy, massive multiplayer online first-person shooter – we can try to generalize some features that allow us to analyze their scalability requisites. Normally, in this kind of games the action takes place in a virtual 3D environment, where thousands of players interact by controlling avatars, allowing real-time interaction between users in simulated virtual worlds. The action environment can be persistent in order to maintain the notion of space and time continuity [wikipedia 2004]. From an analysis of the characteristics of MMOG systems and their usage we can start to identify four main scalability issues: a) simulation capacity that allows for thousands of players to be online in the same virtual world; b) data storage capacity of all the information that is used to represent the virtual worlds and one efficient distribution method for guaranteeing availability when needed; c) reliable and efficient communications for experience coordination and smooth interaction; d) architectural integration enabling system expansibility through new computational, communication and storage resources. Next we will briefly discuss these issues. The simulation component role in MMOG is to process the events that are generated through the player’s interaction or by sub-systems that generate automatic environmental activities (e.g., atmospheric, AI bots). Besides the high event volume that must be processed, the simulation activity has other challenge: the size of the virtual universe data model. Virtual universe action area can have the size of a planet or even a galaxy, which becomes very complex to handle [Rosedale 2003]. As previously referred, the MMOG environments are commonly 3D and very dynamic, being impossible for the clients to keep the virtual world state. So, when a player enters the virtual world must be given to him all the information necessary to animate that world. This information has two different types: data model that represents abstractly the virtual world; and the necessary multimedia elements needed to visual and sonorous animation. Nature and volume size of multimedia information become the main problem of the distribution system [Yu-Shen 1997]. Communication scalability is one of the essential issues in simulated real-time games through Internet. Scalability must be understood not only by the capacity to support communication between a high numbers of players, but also, as the capacity to maintain a communication performance level that doesn’t put at risk the game experience quality. This fact in the MMOG systems is paradigmatic, since there are possible thousand of players interacting with the world objects and moving in the same space. Objects state and players activity must be informed to all players in order to maintain the game integrity/consistency [Smed 2001]. Structural scalability is important to increase the system live span. In order to achieve this requisite the system structure must be designed to enable the addition of new resources. Architectural scalability, through the specification of clear system components that interact in a clear dynamics, through defined protocols, is a pre-requisite for system repairing, actualization and evolution; and must also have the capacity to integrate significantly contribute to the later incorporation of new technologies and devices. At first glance, we would think that to achieve scalability in MMOG implementations we would simply have to work on an architectural design to satisfy all the requisites that have been presented. But that would not be enough. System scalability also emerges from the balance and harmony of the system components. When we are trying to satisfy some requisite, the ideal solution may be in conflict with some other requisite. For example, the best solution for the distribution of static content (such as 3D models, textures and sounds) can jeopardize communication scalability for more immediate real-time events, as their compete for the available bandwidth. The best solution may not be the optimal one for any system component, but the best overall solution for the integrated system, that guarantees an adequate level of quality to the interactive experience. In order to achieve such a balance we have to consider an adequate partitioning of responsibilities for the components and the internal and the external dynamics that are originated.
Contact: Rui Gil, CISUC, University of Coimbra, firstname.lastname@example.org
Copyright is held by the author(s).
Member of collection