FU Berlin, Fachbereich Mathematik und Informatik, Institut für Informatik

Vortrag des Informatik-Kolloquiums

The performance of eCommerce applications is quickly bounded by the database tier. We had to face this issue when evaluating open source middleware from the ObjectWeb consortium such as the JOnAS J2EE application server. As most open source middleware users run their application on commodity hardware they cannot afford large multiprocessor machines and expensive high-end RDBMS. We will present a new concept called Redundand Array of Inexpensive Databases (RAIDb). RAIDb is to databases what RAID is to disks. RAIDb aims at providing better performance and fault tolerance than a single database, at low cost, by combining multiple database instances into an array of databases. Like RAID, we define different RAIDb levels that provide various cost/performance/fault tolerance tradeoffs. RAIDb-0 features full partitioning, RAIDb-1 offers full replication and RAIDb-2 introduces an intermediate solution called partial replication, in which the user can define the degree of replication of each database table. We present a Java implementation of RAIDb called Clustered JDBC or C-JDBC. C-JDBC achieves both database performance scalability and high availability at the middleware level without changing existing applications. We show, using the TPC-W benchmark, that RAIDb-2 can offer better performance scalability (up to 25%) than traditional approaches by allowing fine-grain control on replication.