CVRetrieval: Separating Consistency Retrieval from Consistency Maintenance

Authors

Yijun Lu, University of Nebraska-LincolnFollow
Hong Jiang, University of Nebraska-LincolnFollow
Ying Lu, University of Nebraska-LincolnFollow

Date of this Version

1-4-2007

Comments

University of Nebraska–Lincoln, Computer Science and Engineering
Technical Report TR-UNL-CSE-2007-003
Issued Jan. 4, 2007

Abstract

In distributed online collaboration applications, such as digital white board and online gaming, it is important to guarantee the consistency among participants’ views to make collaboration meaningful. However, maintaining even a relaxed consistency in a distributed environment with a large number of geographically dispersed participants still involves formidable communication and management cost among them.
In this paper, we propose CVRetrieval (Consistency View Retrieval) to solve this scalability problem. Based on the observation that not all participants are equally active or engaged in distributed online collaboration applications, CVRetrieval differentiates the notions of consistency maintenance and consistency retrieval. Here, consistency maintenance implies a protocol that periodically communicates with all participants to maintain a certain consistency level; and consistency retrieval means that passive participants (those with little updating activity) explicitly request a consistent view from the system when the need arises in stead of joining the expensive consistency maintenance protocol all the time. The rationale is that, if a participant does not have updating activities, it is much more cost-effective to satisfy his or her needs on-demand.
The evaluation of CVRetrieval is done in two parts. First, we theoretically analyze the scalability of CVRetrieval and compare it to other consistency maintenance protocols. The analytical result shows that CVRetrieval can greatly reduce communication cost and hence make consistency control more scalable. Second, a prototype of CVRetrieval is developed and deployed on the Planet-Lab test-bed to evaluate its performance. The results show that the active participants experience a short response time at some expense of the passive participants that may encounter a longer response time depends on the system setting. Overall, the retrieval performance is still reasonably high.