Scalable Distributed Data Structures for High-Performance Databases

Present databases, whether on centralized or parallel DBMSs, do not deal well with scalability. We present an architecture for Wintel multicomputers termed AMOS-SDDS, coupling a high-performance main-memory DBMS AMOS-II and a manager of Scalable Distributed Data Structures SDDS-2000. SDDS-2000 provides the scalable data partitioning in distributed RAM, supporting parallel scans with function shipping. AMOS-SDDS potentially allows for scalable data volumes arbitrarily exceeding the storage capacities of a single AMOS-II. We validate our system by experiments with join queries. We show a 30% improvement of response time even for volumes of data that AMOS-II still could manage alone at its site. The results encourage the use of multicomputers for high-performance databases. Key words: Multicomputer, scalability, distributed data structures, parallel processing

• Type: Conference
• Year: 2000
• Where: WDAS
• Authors: Yakham Ndiaye, Aly Wane Diene, Witold Litwin, Tore Risch
• Keywords: Data Structures, Distributed And Parallel Computing, Scalable Data, Scalable Distributed Data, WDAS 2000