Load Balancing for Virtual Worlds by Splitting and Merging Spatial Regions

Umar Farooq, John Glauert, Kashif Zia
2018 Informatica   unpublished
The aggregation algorithm, an integral part of our dynamic infrastructure (using an expansion and a contraction model) for managing scalable virtual worlds, was proposed in our previous work, to overcome the limitations of the current methods using static and hierarchical approaches. The basic aim was to get two contiguous spaces made of smaller regions while distributing the load as balanced as possible among two servers. This algorithm performs well for the perfect square shaped spaces but
more » ... ls when it is applied to spaces of other shapes. The current merging algorithms also assign non-contiguous spaces to servers during the contraction phase. This is due to the unavailability of an explicit continuity check in both aggregation and merging algorithms. In this paper, we provide state-of-the-art in scaling virtual worlds and outline their limitations. It provides both theoretical arguments and simulation results that contiguous spaces have potential benefits. This work, then, extends both the aggregation and merging algorithms and incorporates an explicit continuity check to cope with the issues introduced by allowing non-contiguous spaces. It is demonstrated with the help of results from our prototype that the extended methods strictly achieves the theoretical goals of the proposed methods. Povzetek: Podan je pregled skalirnih metod v navideznih svetovih (VWs) in nov algoritem za razširjanje in krčenje podprostorov.
fatcat:lanysbzm45czlpkk2ugnwdq7na