A Protocol for Reconciling Recovery and High-Availability in Replicated Databases

Abstract

Abstract. We describe a recovery protocol which boosts availability, fault tolerance and performance by enabling failed network nodes to resume an active role immediately after they start recovering. The protocol is designed to work in tandem with middleware-based eager update-everywhere strategies and related group communication systems. The latter provide view synchrony, i.e., knowledge about currently reachable nodes and about the status of messages delivered by faulty and alive nodes. That enables a fast replay of missed updates which defines dynamic database recovery partition. Thus, speeding up the recovery of failed nodes which, together with the rest of the network, may seamlessly continue to process transactions even before their recovery has completed. We specify the protocol in terms of the procedures executed with every message and event of interest and outline a correctness proof.

Optimizing recovery protocols for replicated database systems

“…Based on the proposal made in [2], we review the functionality of the original recovery protocol and enhance it incorporating a compacting method for improving its performance and providing an accurate amnesia support. To this last, we consider some related works [21,30,23], and I want to thank to Rubén de Juan Marín for his extensive and valuable contribution to this work by providing the foundation basis for analysis and incorporation of a solution to the amnesia phenomenon in the recovery protocol implemented.…”

“…Our basic proposal is inspired in the recovery protocol presented in [2]. It has been designed for eager update everywhere database replication protocols and proposes the use of DB-partitions (see below).…”

“…It has been designed for eager update everywhere database replication protocols and proposes the use of DB-partitions (see below). It was originally designed for providing recovery support for the ERP and TORPE [2] replication protocols. Such protocols use a voting termination approach [60], and can be considered as weak voting replication protocols [63].…”

“…As we have briefly highlighted before, these replication protocols provide high availability. However, only a few of them deal with the possible reconnection of the failed node, which is managed by recovery protocols [12,39,43,2].…”

“…This chapter is focused in the recovery protocol for eager update everywhere replication protocols, proposing one optimization to the work presented in [2]. This enhancements includes amnesia support [20] and a better performance reducing the amount of data to save in the actions done before recovering and the amount of data to transfer at recovering time.…”