By Topic

Uniform and Non-Uniform Zoning for Load Balancing in Virtual Environments

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$33 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

2 Author(s)
Dewan Tanvir Ahmed ; Distrib. & Collaborative Virtual Environments Res. Lab., Univ. of Ottawa, Ottawa, ON, Canada ; Shervin Shirmohammadi

Maintaining a stable system for thousands of users is a difficult task. Massively Multiplayer Online Game (MMOG) dealing thousands of players preserves consistency by sharing relevant game states among the interacting parties. The lag of state sharing becomes excessive when the system is overloaded. Current practices supporting a massive number of users generally divide the game world into zones which are managed by servers. However, such zoning restricts cross-zonal interactions and exposes division of the game space. To address these problems, we present load-balancing algorithms for both uniform and non-uniform zonal Peer-to-Peer (P2P) MMOGs. The proposed load-balancing schemes identify a loaded server in terms of either the number of players or packets processed per unit time, and then move the load to other servers considering communication overhead and P2P overlay restructuring. The non-uniform load balancing, named adaptive scheme, uses a bisection procedure that does not adhere to any predefined zone size - zone sizes are flexible and can be determined dynamically. The outlined Multilevel Multiphase Load Balancing (MMLB) method designed for uniform zones reduces load in a step-by-step manner, and avoids problems associated with current load balancing schemes. Our results show that reducing load at any magnitude not necessarily improves performance. By comparison, MMLB performs better than adaptive scheme especially for a series of hotspots.

Published in:

2010 5th International Conference on Embedded and Multimedia Computing

Date of Conference:

11-13 Aug. 2010