Dr. Raja Appuswamy

Research

Hardware acceleration for data-intensive domains

Molecular information storage

Synthetic DNA has the potential to emerge as an ideal archival storage media--it is extremely dense, highly durable, has eternal relevance, consumes no power. In projects Oligoarchive and MOSS, we investigate the problem of using DNA as a reliable storage medium for archiving digital data.

Cold data management

Enterprises continues to accumulate vast amounts of information driven by the desire to extract actionable insights. However, not all data is frequently accessed. Recent studies claim that as much as 80% of this data is cold, or infrequently accessed. Cost-effective storage of this cold data has become very important to minimize the overall cost of data analytics. Our research focuses on several aspects of cold data management, from economic modeling of storage costs, to query execution techniques for high-latency, nearline storage devices like Virtual Tape Libraries.

Heterogeneous, parallel database engines

Modern real-time reporting applications require the ability to perform real-time analytics over fresh operational data without compromising transactional throughput. However modern database engines struggle to meet such conflicting demands as are limited to CPUs. The Caldera project focuses on combining the data-parallel nature of GPUs and task-parallel nature of CPUs in a synergistic fashion for building new Heterogeneous, hybrid transactional and analytical processing database engines.