DSL Research Areas

• Query Optimization:
  
  Design of techniques for effectively reusing plans produced by the query optimizer for earlier queries to generate the plans for fresh queries.
  Development of tools for characterizing the behavior of industrial-strength database query optimizers.
  Improving the speed and scaling of query optimizers.
  Papers in VLDB 2002, SIGMOD 2003, VLDB 2004, VLDB 2005, ICDE 2007, VLDB 2007, VLDB 2008, VLDB 2010.
  
  
• Biological Databases:
  
  Development of index structures and buffer management policies for efficiently supporting genomic (DNA and protein) sequences. Development of database systems for hosting bio-diversity information.
  Papers in COMAD 2000, Information Systems Journal 2003, ICDE 2004, SIGMOD 2004, HiPC 2005.
  
  
  

In the Recent Past

• Multilingual Databases:
  
  Characterization of the performance impacts of storing data in multilingual format. Design of database architectures that address multilingual performance drawbacks. Design of multilingual SQL operators.
  Papers in VLDB 2003, ICDE 2004, EDBT 2004, SIGMOD 2004, CIKM 2004, DASFAA 2005, ICDE 2006.
  
  
• Privacy-Preserving Mining:
  
  Development of privacy-preserving techniques for extraction of association rules on large boolean databases.
  Papers in VLDB 2002, DASFAA 2004, ICDE 2005, DMKD 2009.
  
  
• XML Databases:
  
  Design of translation mechanisms for storing XML documents in relational engines. Design of XML compression techniques that directly support querying in the compressed domain. Design of XML indexing systems.
  Papers in ICDE 2002, SIGMOD 2002, VLDB 2002, ICDE 2003, ICDE 2005, COMAD 2005b, DASFAA 2006, Information Systems Journal 2007.
  
  
• Web Databases:
  
  Design of server-side cacheing mechanisms for dynamic Web content.
  Papers in SIGMOD 2002, APWEB 2004, COMAD 2005b.
  
  
• Database Mining:
  
  Design of efficient, incremental and concise data mining algorithms for boolean association rules.
  Papers in Information Systems Journal 2000, SIGMOD 2000, ICDE 2002, PAKDD 2002, ICDE 2003, PAKDD 2003.
  
  
  

In the Previous Millenium

• Distributed Web Database Processing:
  
  Designed a new Web querying system called Discover that processes queries in a truly distributed manner, thereby deriving several advantages, whereas most other systems proposed in the literature have a centralized processing architecture.
  
  
• Database Compression:
  
  Data compression has traditionally been used for reducing disk space usage in database systems. A major problem, however, is that processing queries on compressed databases becomes very slow. We have developed novel compression algorithms that simultaneously provide fast query processing and significant disk savings.
  
  
• Transaction Management in Distributed Databases:
  
  Incorporating distributed data into the database framework incurs the well-known additional complexities that are associated with transaction concurrency control and recovery. We have designed high-performance distributed concurrency control protocols and distributed commit protocols.
  
  
• Object-oriented Database Solutions for Advanced Applications:
  
  Modern applications place sophisticated demands on the functionality expected from database systems. Solutions based on object-oriented database systems are required to adequately support such applications. The applications that we have built using OO technology include Flexible Manufacturing, VLSI Design and Bio-Diversity.
  
  
• Resource Management Algorithms for Real-Time Database Systems:
  
  In real-time database applications, transactions have completion deadlines and the system performance is measured in terms of missed deadlines, not response time. Therefore, new database resource management algorithms have to be developed to efficiently support the new metric. Our work includes concurrency control protocols, index access protocols, recovery protocols and security protocols.
  
  
• Performance Guarantees in Overloaded Real-Time Systems:
  
  Real-time systems may encounter emergency situations wherein the processing required to handle the emergency exceeds the system capacity, resulting in system overload and disastrous consequences. We have designed real-time scheduling algorithms that provably provide a guaranteed level of performance even under overload conditions.
  
  
• Scheduling in I/O Subsystems:
  
  There is an increasing demand on computer systems to provide service guarantees to end-users with regard to a variety of metrics. Multimedia applications, for example, require meeting real-time and synchronization constraints. Time-shared systems, on the other hand, are usually expected to provide fair service. We have studied this issue in detail for I/O subsystems and developed a new set of disk scheduling algorithms that efficiently achieve the desired goals.