Scalable entity-centric prediction based concurrency control for distributed virtual environments

Abstract

Replication is often used to allow high performance interactive operations to the users of distributed virtual environments. This is important especially when users are geographically distributed over large networks like the Internet. On the other hand, multiple concurrent updates due to replication may lead to inconsistent views among users. This results in the need of concurrency control. Concurrency control schemes used in the existing systems are broadly categorized into pessimistic, optimistic and prediction schemes. The pessimistic scheme interferes with users in natural interaction due to communication delay while the optimistic scheme confuses users due to repairs. The prediction scheme provides acceptable interactive performance and consistency. However, the existing prediction scheme does not scale as the number of users increases. This paper proposes a prediction based concurrency control scheme that satisfies the needs for large scale distributed virtual environments in terms of interactive performance as the number of users increases. The proposed scheme allows real time interactions for users as the optimistic concurrency control scheme and eliminates the need for repairs as the pessimistic concurrency control scheme. We exploit the entity-centric multicast: only the users surrounding a target entity do multicast the ownership requests for ownership prediction by joining the multicast group address assigned to the entity. We have implemented the proposed scheme as a library and evaluated its performance using a test-bed application, CVRAT. Also we have performed to a set of experiments to demonstrate the efficiency and scalability of the proposed scheme. The first set of experiments was run to find an optimal geometry size of an entity-centric multicast group. We have learned from the results that the size must be at least as large as the twofold of latency. The second set of experiments was designed to demonstrate the scalability of the propo...