Scaling virtual worlds: Simulation requirements and challenges

Abstract

Virtual worlds use simulation to create a fully-immersive 3D space in which users interact and collaborate in real time. It is still a great challenge to scale virtual worlds to provide rich user experiences, high level of realism, and innovative usages. There are three unique simulation requirements in scaling virtual worlds: (1) large-scale, real time and perpetual simulations with distributed interaction, (2) simultaneous visualization for many endpoints with unique perspectives, and (3) multiple simulation engines with different operation characteristics. In this paper, we review the challenges in meeting these requirements, present the scalability barriers we observed in current virtual worlds, and discuss potential virtual world architecture and solutions to address the challenges and overcome the barriers. INTRODUCTIONVirtual worlds use simulation to create a fully-immersive 3D space for users to explore, collaborate, and interact in real time. There are two broad classes of popular virtual worlds: general purpose virtual worlds, such as Second Life®, and massively multiplayer online games, such as World of Warcraft™. Scalability, however, is still a great challenge in virtual worlds. For instance, Second Life and World of Warcraft are limited to below 100 users interacting with each other concurrently (Gupta et al. 2009). To meet the increasing demand of rich user experiences, high level of realism, and new usages such as experiencing professional sports games in virtual auditoriums, virtual worlds must scale beyond their current capability in several dimensions.The first dimension is to scale the number of concurrent users interacting with each other. Current virtual worlds have to split the user base and restrict interactions to achieve scalability. For instance, Second Life applies static space partitioning to decompose the space into 256m x 256m regions, each handled by one server. World of Warcraft uses sharding: a part of the virtual world is replicated into shards and different shards reside on different servers, but users on different shards are isolated from each other. Both approaches degrade the massive multi-user experience: sharding prevents large groups of users from interacting by design and static partitioned regions collapse with too many users (Gupta et al. 2009).The second dimension is to scale the scene complexity, captured by the number of objects and the complexity of their behaviors and appearance. It is especially challenging in general purpose virtual worlds, where user generated content is dominant. Designers of such a world only provide tools, basic parts, and primitive objects (prims) to enable users to be creative and produce content. There is no limitation on what users can build and potentially infinite set of object behaviors could be expressed, hence the term "general purpose". As such, the tools and prims are not optimized for any special purposes. When a 778 978-1-4244-9865-9/10/$26.00 ©2010 IEEE

3D Virtual worlds and the metaverse

“…For this discussion, "real-time" refers to the simplified metric of frame rate, or the frequency with which a computer graphics system can render a scene. As graphics hardware and algorithms have pushed the threshold of what can be computed and displayed at 30 or more frames per second-a typically accepted minimum frame rate for real-time perception [Kumar et al 2008;Liu et al 2010]-virtual world applications have provided increasingly detailed visuals commensurate with the techniques of the time. Thus, alongside games, 3D modeling software, and to a certain extent 3D cinematic animation (qualified as such because the real-time constraint is lifted for the final product), virtual worlds have seen a progression of visual detail from flat polygons to smooth shading and texture mapping and finally to programmable shaders, which can apply transformations and other computations to graphical elements with great efficiency and flexibility.…”

“…Based on this article's prime criterion that a fully-realized Metaverse must provide a milieu for human culture and interaction, scalability may thus be the most challenging virtual world feature of all, as the physical world is of enormous and potentially infinite scale on many levels and dimensions. Three dimensions of virtual world scalability have been identified in the literature [Liu et al 2010]. …”

“…However, as pointed out by Ian Wilkes and borne out by subsequent empirical studies [Gupta et al 2009;Liu et al 2010], partitioning by region fails on key virtual world use cases.…”

“…OpenSimulator, as a reverse-engineered port of the Second Life architecture, thus also manifests the same strengths and weaknesses of that architecture [OpenSimulator Project 2011]. It has, however, also served as a basis for more advanced work on virtual world scalability [Liu et al 2010]. …”

“…The distributed scene graph (DSG) approach separates the state of the virtual world into multiple scenes, with computational actors each focused on specific types of changes to a scene (e.g., physics, scripts, collision detection) and communicationintensive client managers serving as liaisons between connected clients/viewers and the virtual world [Liu et al 2010]. This separation strategy isolates distinct computational concerns from the data/state of the world and communications among clients that are subscribed to that world.…”