Understanding replication in databases and distributed systems

Abstract

Replication is an area of interest to both distributed systems and databases. The solutions developed from these two perspectives are conceptually similar but differ in many aspects: model, assumptions, mechanisms, guarantees provided, and implementation. In this paper, we provide an abstract and "neutral" framework to compare replication techniques from both communities in spite of the many subtle differences. The framework has been designed to emphasize the role played by different mechanisms and to facilitate comparisons. With this, it is possible to get a functional comparison of many ideas that is valuable for both didactic and practical purposes. The paper describes the replication techniques used in both communities, compares them, and points out ways in which they can be integrated to arrive to better, more robust replication protocols.

Pronto: High availability for standard off-the-shelf databases

“…More recently, several works have proposed implementing eager update everywhere replication using an underlying total order broadcast abstraction (e.g., [4,14,25,41,55]). In general, these protocols capture different trade-offs between scalability (in terms of throughput) of the replicated database system, underlying network assumptions, generality of the transaction model, and use of standard or custom databases.…”

Untitled

“…It is only recently that atomic broadcast has been considered as a possible candidate to support replication, as termination protocols (see [35] and [36] for brief surveys on the subject). Schiper and Raynal [29] pointed out some similarities between the properties of atomic broadcast and static transactions (transactions whose operations are known in advance).…”

Reviewing Amnesia Support in Database Recovery Protocols

“…All of them are proposed for database replication protocols sharing the following characteristics: update everywhere protocol -ROWAA approach [16]-, based on total order broadcast -without a terminating phase-propagating a message per transaction -constant interaction [28]-and virtual synchrony. Moreover, replicated data objects are tagged with version numbers.…”

“…Latest trends in full database replication techniques -managed by replication protocols [24,25,23,26,13,2]-make use of a Group Communication System (GCS for short) [9] as it is detailed in [28]. These GCSs offer different services to the systems built atop of them.…”