Summary of the Project

Introduction

An in-memory database is a database that runs entirely in main memory, without incurring an I/O from/to disk, and the data structures and access algorithms exploit this property to achieve breakthrough performance. When there is enough memory available, an in-memory database is able to provide significant performance improvement in response time compared to a disk-based database system.

The primary goals of the project are:

i)      To do a feasibility study analysis, of using in-memory index structures such as BitCube and Quasi-BitCube and the Oracle TimesTen in-memory database, on the SGI 4700 4.1 terabyte main memory system at LITE, to identify bottlenecks and subsequently propose solutions.

ii)     Use the analysis results to drive/propose a new model that could lay the foundations for the design and development of the next generation database management system.

Through this project we seek to create unparalleled economic opportunity by engaging businesses (e.g. Crossroads Systems, SGI, Oracle, etc.) at various stages of the innovation process: needs & gaps identification, plan formulation, research, development and technology transfer.

Sub-Projects

T-Tree

 project workspace (access restricted)

MX-Tree