Convergent Causal Consistency for Social Media Posts
Ta-Yuan Hsu, Ajay D. Kshemkalyani · 2021
Geo-replicated services play a vital role in cloud storage management by providing enhanced availability, higher reliability, and lower latency on demand access to shared infrastructure and data resources. In such environments, consistency is a critical and essential consideration for distributed storage systems where it is required to make updates to the replicated data. Convergent causal consistency has become a popular consistency model by offering useful semantics for online human interaction services. Adaptive non-full replication strategies have potential benefit of lower message counts in social network systems. However, static replication is ineffective for time-varying workloads. This paper presents a causal+ consistency protocol, CaDRoP, to support adaptive dynamic replication with the convergence property for all comments following a post and the causal ordering between posts with explicit causality. We evaluate CaDRoP protocol with realistic workloads by different PUT rates in terms of the practical price of Amazon Web Service. The results show that CaDRoP can yield significantly lower cost than it is running in a statically replicated data store. We further evaluate CaDRoP by comparing it with a clairvoyant optimal replication solution. The findings indicate that with cache, CaDRoP incurs only around 6% ~ 16% extra cost. Without cache, CaDRoP brings around 2% ~ 4.5% extra cost in steady states.