Service of SURF
© 2025 SURF
There still remains many challenges in creating an effective and efficient cloud gaming operation able to handle the new release games of today. There are also many existing paths to creating a cloud gaming architecture. In this paper some of the different approaches, that are beyond that of just remote rendering, are analysed giving insight into the operational approach of each technology. Currently there is a growing number of initiatives in cloud game architectures that vary in significant ways. Although there are many varying technologies, with a lot of promises, the ultimate goal of a cloud game engine is something unique to what has been before. It really is about providing a modular and scalable approach but within a controllable and sandbox like environment.
Offering games in the cloud is something being realised by many operations, but still there are diverse challenges in conducting this successfully. Currently it is expected that graduating game development students will have at least one published game and a chance to experience the release phase and post-production aspects of game development. This led to the inspiration for a project to create the first University cloud games channel in an effort to stop our game development history being lost due to reasons such as storage media failure, along with providing an avenue for students to experience game release, operation and community management. A study evaluated various technologies and requirements from stake-holders. The criteria created was then used to identify the best approach to undertake and implement. The results includes the selection and development of the University games channel using an asset streaming solution in partnership with Uiomik. along with the supporting processes for games to be delivered to the platform. This solution makes use of specialised offline asset compression combined with asset streaming prediction. The game requires no installation and the performance of the caching of the game required to start playing was profiled and found to perform at an average of 29 percent of the total game size for our student games, that was then compared to the results of industry games that performed on average at 4 percent.