A Proactive, Cost-aware, Optimized Data Replication Strategy in Geo-distributed Cloud Datastores

Geo-replicated cloud datastores adopt the replication methodology by placing multiple data replicas at suitable storage zones. This can provide reliable services to customers with high availability, low access latency, low system cost, and decreased bandwidth consumption. However, this has the potential to increase the whole system overheads of maintaining more resource replicas, and to also degrade the system utilization due to unnecessary storage space cost. Thus, it is important to determine the suitable replication zones on-the-fly to increase the availability of data resources and maximize the system utilization. Specifically, it is essential to determine the appropriate number of replicas for different data resources at each zone in a particular time interval. We propose Cost Optimization Replica Placement (CORP) algorithms to enable state-of-art proactive provisioning replication of data resources based on an one-step look-ahead workload behavior pattern forecast over the distributed data storage infrastructure using statistical techniques. The experimental results show the cost effectiveness of the proposed replication strategies.