This thesis describes Vivaldi, a self-configuring architecture for scaling the database tier of dynamic content web server clusters. We design an automatic and transparent provisioning algorithm that provides a unified approach to load and fault management in a database cluster. Our algorithm is based on dynamic data replication and provides strong consistency guarantees (i.e., 1-copy serializability) for the database cluster. Our algorithm dynamically allocates tasks to commodity databases across workloads in response to peak loads or failure conditions thus providing quality of service to clients.; We investigate our transparent provisioning mechanisms in the database tier using two dynamic content applications: the TPC-W e-commerce benchmark and the RUBIS auction site benchmark. We demonstrate that our techniques provide both high resource usage and quality of service under different load and failure scenarios.
展开▼
