GORDA: An Open Architecture for Database Replication

Abstract

Recently, third party solutions for database replication have been enjoying an increasing popularity. Such proposals address a diversity of user requirements, namely preventing conflicting updates without the overhead of synchronous replication; clustering for scalability and availability; and heterogeneous replicas for specialized queries. Unfortunately, the lack of native support from database vendors for third party replication forces implementors to either modify the database server, restricting portability, or to develop a middleware wrapper, which causes a performance overhead. This paper addresses this problem with a novel architecture and programming interface for replication, such that different strategies can be efficiently implemented on * Parts of this extended report were published in the Proceedings of the 6th IEEE Internacional Symposium on Network Computing and Applications (NCA '07), Boston, MA, USA. 2007. any compliant database management system in a cost-effective manner. The contribution is twofold. First we propose a reflective model of transaction processing and explain how it can be used to achieve replication. Then we implement the proposed architecture in Apache Derby, PostgreSQL, and Sequoia and evaluate the PostgreSQL implementation with the TPC-W industry standard benchmark.

Improving the Scalability of Cloud-Based Resilient Database Servers

“…Handlers, or hooks, are assumed to exist and are called by the DBMS Transaction Manager. A set of interfaces targeting this behavior has been proposed and several prototypes exist [14]. Nevertheless, these hooks are further explained in the next few lines.…”

Modelling adaptive services for distributed systems

“…For illustration purpose, we shall use a middleware solution for database replication based in the results of the GOR-DA project [4]. The system is composed by a set of database servers.…”

“…For that purpose, the replication uses a group communication service implementing totally ordered atomic broadcast [12]. A description of the database consistency algorithm is outside the scope of this paper (the interested reader may refer to [4]). The service composition of the Data channel may include an optional auditing service, that can be present at each node for performance monitoring and management.…”

Performance evaluation of database replication systems

“…Failures of computers are masked by replicating the data in several computers. Many protocols have been proposed in the literature targeting different consistency and scalability guarantees [19,22,26,15,4,5,14,25,12,13,24,17,21,20,7].…”