Distributed Information Systems

We work at the intersection of data management and distributed systems to develop the next generation of distributed information systems. We design algorithms, build prototype systems, and transfer technology to IBM's software products.

In the past, our group did pioneering research on parallel database systems, which has provided underlying technology for IBM DB2's widely deployed data sharing product. We also developed DataLinks technology for DB2, and have more recently collaborated on autonomic computing for DBMSs.


Current Projects

• Scalable Database Architectures: We are investigating new architectures for database systems, along two dimensions: (a) large-scale parallelism over non-dedicated computers and clusters, (b) and embedded data processing functionality in storage devices. This is in collaboration with IBM's storage research group.
• Dynamic Federation: This project is developing a distributed data management middleware for flexible querying of large scale, autonomous data sources. The middleware includes a data placement advisor for automatically analyzing query patterns and forming appropriate replicas to meet QoS needs, a metawrapper for dynamic federation of autonomous data sources, and an autonomous indexing scheme based on microeconomic principles.
• Autonomic Replication Management Service (ARMS): ARMS is a web services based common management framework for easy deployment, configuration and management of data replication products in an on demand operating environment.
• Stream Data Management: SIRIUS is a distributed stream data management system that supports event algebra and relational algebras. It investigates data modelling, support for continuous queries and adaptive query processing in the context of event streams.
• Entropy Compression of Relations: With modern processors, data movement costs (cache-memory-disk) often dwarf computation costs. To address this imbalance, we are investigating the compressibility of relations, and algorithms for querying compressed relations. Our initial results suggest that we can compress relations down to their entropy, gaining about an order of magnitude reduction in program working sets, while still being able to query efficiently.

• Relational Programming Language  (V. Raman). Working Draft.
• Progressive Query Optimization for Federated Queries  (S. Ewen, H. Kache, V. Raman, and V. Markl). International Conf. on Extending Data Base Technology (EDBT), 2006. (to appear)
• Adaptive Query Processing  (A. Deshpande, V. Raman). Intl Conf on Management of Data (COMAD), 2005. Tutorial.
• Evolving Toward the Perfect Schedule: Co-scheduling Job Assignments and Data Replication in Wide-Area Systems Using a Genetic Algorithm. (T. Phan, K. Ranganathan, and R. Sion) Proceedings of the Workshop on Job Scheduling Strategies for Parallel Processing (JSSPP), 2005.
• Trading Off Security in a Service Oriented Architecture. (Garret Swart, Benjamin Aziz, Simon N. Foley, John Herbert) DBSec 2005
• More

• Vijay Dialani
• Dengfeng Gao
• Vitthal Gogate
• Wei Han
• Wen-Syan Li
• Thomas Phan
• Vijayshankar Raman
• Garret Swart

• David McWherter (CMU)
• Prof. Norman Paton (U. Manchester)
• Yan Gu (Georgia Tech)
• Prof. K. Selcuk Candan (Arizona State)
• Yogesh Simmhan (U. Indiana)
• Kavitha Ranganathan (U. Chicago)
• Shawn Jeffery (U. Wisconsin)
• Jon McAlister (Stanford U.)