Storage and network resource usage in reactive and proactive replicated storage systems

Replicated storage systems allow their stored data to outlive the life of the nodes storing them by use of replication. In such systems, failed replicas are “repaired” by copying remaining replicas from other nodes. We consider two main replication strategies: reactive, in which replication occurs in response to failures, and proactive, in which replication occurs in anticipation of failures. There is no consensus in the literature about which strategy is better. Our work presents a quantitative analysis that compares reactive and proactive through simulations and verified analytically where possible. In particular, we systematically compare how the strategy of replication affects the usage of storage and network resources, at peak consumption as well as on average, to achieve a given lifetime for a replicated object. Our results indicate that a proactive strategy leads to multiple times higher storage requirements than a reactive strategy, with the exact number depending on parameter values for failure and replication rates. Reactive systems are moderately bursty in terms of bandwidth consumption, with rare peaks of at most five times (for realistic parameter values) that of proactive systems, the latter of which is constant by design given that replications are periodic for proactive.