TDBRepl - The Trivial DataBase Replication System

Introduction

The Trivial Database Replication System is an open source suite of 
software that provides a high-performance, replicated, distributed Unix 
DBM style database built on Andrew Tridgell's TDB. The software allows
UNIX systems to utilise a distributed, replicated peer-to-peer database.
The peer-to-peer design means that there are no single points of failure
such as a 'master' server used in most replication models. Requests to
retreive data are served by a database copy on local disk while 
requests to delete/add/modify data are sent over a TCP/IP network.

The system was created to furfil the following requirements:

	- small, lightweight DBM style database (no SQL);
	- no single point of failure (e.g master/slave configuration);
	- ability to function without full complement of nodes;
	- ability to replay transactions automatically to nodes that
	  were unreachable for a period;
	- simple drop-in replacement for TDB/DBM API;
	- support multiple (thread and process) writers on the same database;

Home

Last update: Thursday, 22-May-2008 01:25:39 UTC