Date of this Version
In Internet-scale distributed systems, replication-based scheme has been widely deployed to increase the availability and efficiency of services. Hence, consistency maintenance among replicas becomes an important research issue because poor consistency results in poor QoS or even monetary loss. Recent research in this area focuses on enforcing a certain consistency level, instead of perfect consistency, to strike a balance between consistency guarantee and system’s scalability.
In this paper, we argue that, besides balancing consistency and scalability, it is equally, if not more, important to achieve adaptability of consistency maintenance. I.e., the system adjusts its consistency level on the fly to suit applications’ ongoing need. This paper then presents the design, implementation, and evaluation of IDEA (an Infrastructure for DEtection-based Adaptive consistency control), which adaptively controls consistency in replicated services by utilizing an inconsistency detection framework that detects inconsistency among nodes in a timely manner. Besides, IDEA achieves high performance of inconsistency resolution in terms of resolution delay.
Through two emulated distribution application on Planet-Lab, IDEA is evaluated from two aspects: its adaptive interface and its performance of inconsistency resolution. According the experimentation, IDEA achieves adaptability by adjusting the consistency level according to users’ preference on-demand. As for performance, IDEA achieves low inconsistency resolution delay and communication cost.