Please use this identifier to cite or link to this item:
|Title:||Multi-server support multiplayer online games|
|Authors:||Wong, Biu Pui|
|Department:||Department of Computer Science|
|Supervisor:||Dr. Lau, Rynson. First Reader: Dr. Komura, Taku. Second Reader: Dr. Jia, Xiaohua|
|Abstract:||Collaborative Virtual Environments (VE) is very popular especially for entertainment purposes nowadays in the online games market. The VE is becoming more complex and huge, and the amount of concurrent users is increasing. As there is always higher scalability, processing power and memory resources in the multiple servers system, thus most of the VE were managed by multiple servers together rather than a single server. However, the real-time interaction between the concurrent users and the VE, the performance of the VE and the system resources which is needed for maintaining the VE are threatened by the overloading problem still. Unlimited to append more additional servers to avoid the problem of overloading is always impossible in the business world because of the cost-effective reasons. In order to hold and maintain the VE more cost-effectively, it is necessary to avoid overloading occur in any one of the servers. Thus, a better load-balancing mechanism is vital for this kind of system. A lot of related solutions for load-balancing from various past research from many scholars existed already. However, some of the solutions can only balance the workload locally, hence cannot effectively to evenly redistribute the workload under the condition of highly-skewed unevenly distributed of workload. The other solutions only perform load balancing globally which always costly for processing power, memory resources and network resources under any conditions even though when the situation is not necessary to solve the overloading problem globally. And the existing scalable schemes takes longer duration to determined all parameters in advance to for a scalable load balancing action before perform workload reallocation, and never consider the real time changes of distribution virtual environment (DVE), thus not real time enough to take action and not flexible to deal with the real time changing DVE. Hence, a more cost-effective and practicability scalable solution which also more flexible to adapt the real time changing DVE will be beneficial for those multi-server system for supporting DVE. Thus, this project is going to analysis some existing solutions, find out their limitations, and deliver a load-balancing solution which is including some improvements for the limitations of some existing solutions and more flexible to adapt the real time changing DVE. In order to deliver a load-balancing solution with improvements, an effective locally load-balancing solution and an effective scalable load-balancing solution will be selected and analyzed in order to find out their merits to apply to the solution of this project and overcome their own limitations. The selected existing solutions will be implemented for testing their practicability and effectiveness. Then a solution for load-balancing will be designed according to the design of the selected solutions and analysis. In order to prove the practicability, effectiveness and efficiency of the design of the load-balancing solution of this project, the design will be implemented for experiment. Finally, evaluation and analysis for the solution of this project will be done for evaluate the value of the more comprehensive solution delivered. A load-balancing solution with some improvements is anticipated to be delivered. It is anticipated as a cost-effective solution which can solve the overloaded situation locally or more-globally according to different cases of overloading and different workload distribution pattern within a DVE. The solution is anticipated capable to perform load balancing action more real time and more flexible to adapt the real time changes of DVE. Then the effectiveness of the delivered load-balancing mechanism is anticipated to be raised. And the workload among the server is anticipated to be more evenly distributed. Neither games nor actual VE which do not help for achieving the project goals will be delivered. As the limitation of hardware resources, some status of system and VE will be simulated or asserted. This project cannot guarantee the original performance can be remained, as the past approaches by various scholars may be implemented by different technology and work under various environments. All the deliverable can only work under Microsoft Windows platform as they will be implemented as Windows application for experiments. The consistence of the results of experiments will not be guaranteed as the network conditions usually unanticipated. The delivered load-balancing solution is anticipated may be beneficial for developing a more effective and efficiency multiple servers system for holding the DVE and deal with huge amount of concurrent users.|
|Appears in Collections:||Computer Science - Undergraduate Final Year Projects|
Items in CityU IR are protected by copyright, with all rights reserved, unless otherwise indicated.