To Infinity and Beyond: Scaling Massively Multiplayer Games
Authors
Cohen, Ron ; Ejlersen, Anders ; Kristensen, Rasmus
Term
2. term
Education
Publication year
2010
Submitted on
2010-05-31
Pages
104
Abstract
Massivt multiplayer-online-spil (MMOG'er) kræver lastbalancering—at fordele arbejdet mellem servere—for at holde spillet kørende uden hak. Vi undersøger geografisk partitionering, som deler spilverdenen op i områder, der håndteres af forskellige servere. Vi analyserer tre eksisterende todimensionelle metoder (statisk, matrix og JoHNUM-partitionering) og udvider dem til tredimensionelle virtuelle miljøer. For at teste udvidelserne byggede vi et enkelt testspil, Rock Pounder, og vurderede metoderne i to scenarier, der afspejler typisk MMO-spil. Resultaterne viser, at statisk partitionering passer til stabile og forudsigelige belastninger. Ved skiftende, dynamiske belastninger klarer den dynamiske JoHNUM-tilgang sig bedst, især fordi den laver en mere grundig analyse af verdenen før opdeling. Samlet set viser JoHNUM ret god skalerbarhed i forhold til antallet af opererende servere.
Massively multiplayer online games (MMOGs) rely on load balancing—spreading work across servers—to keep play smooth. We study geographic partitioning, which divides the game world into regions managed by different servers. We analyze three existing two-dimensional methods (Static, Matrix, and JoHNUM partitioning) and extend them to three-dimensional virtual environments. To test these extensions, we built a simple game, Rock Pounder, and evaluated the methods in two scenarios that reflect common MMO play. Our results show that static partitioning suits stable, predictable workloads. For changing, dynamic workloads, the dynamic JoHNUM approach performs best, largely because it includes a more thorough analysis of the world before partitioning. Overall, JoHNUM shows fairly good scalability relative to the number of operating servers.
[This abstract was generated with the help of AI]
Documents