David J. DeWitt
Research Subjects
David J. DeWitt has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Qi Su
University of Wisconsin-Madison
2 shared publications - Ravishankar Ramamurthy
University of Wisconsin-Madison
2 shared publications - Daniel J. Abadi
Computer Science and Artificial Intelligence Lab. Massachusetts Institute of Technology
1 shared publication - Daniel S. Myers1 shared publication
- Eric Robinson
University of Wisconsin -Madison 1210 W. Dayton St. Madison 53706 WI
1 shared publication - Manish Mehta
12, S. First St., San Jose, CA 95113, USA
1 shared publication
Latest Publications
- 1 April 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 10636382,p. 466-475
Materialization Strategies in a Column-Oriented DBMS
There has been renewed interest in column-oriented database architectures in recent years. For read-mostly query workloads such as those found in data warehouse and decision support applications, "column-stores" have been shown to perform particularly well relative to "row-stores" In order for column-stores to be readily adopted as a replacement for row-stores, however, they must present the same interface to client applications as do row stores, which implies that they must output row-store-style tuples. Thus, the input columns stored on disk must be converted to rows at some point in the query plan, but the optimal point at which to do the conversion is not obvious. This problem can be considered as the opposite of the projection problem in row-store systems: while row-stores need to determine where in query plans to place projection operators to make tuples narrower, column-stores need to determine when to combine single-column projections into wider tuples. This paper describes a variety of strategies for tuple construction and intermediate result representations and provides a systematic evaluation of these strategies.
- 1 April 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 10636382,p. 466-475
Materialization Strategies in a Column-Oriented DBMS
There has been renewed interest in column-oriented database architectures in recent years. For read-mostly query workloads such as those found in data warehouse and decision support applications, "column-stores" have been shown to perform particularly well relative to "row-stores" In order for column-stores to be readily adopted as a replacement for row-stores, however, they must present the same interface to client applications as do row stores, which implies that they must output row-store-style tuples. Thus, the input columns stored on disk must be converted to rows at some point in the query plan, but the optimal point at which to do the conversion is not obvious. This problem can be considered as the opposite of the projection problem in row-store systems: while row-stores need to determine where in query plans to place projection operators to make tuples narrower, column-stores need to determine when to combine single-column projections into wider tuples. This paper describes a variety of strategies for tuple construction and intermediate result representations and provides a systematic evaluation of these strategies.
- Open Access
Turning Cluster Management into Data Management: A System Overview
This paper introduces the CondorJ2 cluster management system. Traditionally, cluster management systems such as Condor employ a process-oriented approach with little or no use of modern database system technology. In contrast, CondorJ2 employs a data-centric, 3-tier web-application architecture for all system functions (e.g., job submission, monitoring and scheduling; node configuration, monitoring and management, etc.) except for job execution. Employing a data-oriented approach allows the core challenge (i.e., managing and coordinating a large set of distributed computing resources) to be transformed from a relatively low-level systems problem into a more abstract, higher-level data management problem. Preliminary results suggest that CondorJ2's use of standard 3-tier software represents a significant step forward to the design and implementation of large clusters (1,000 to 10,000 nodes).
Open AccessTurning Cluster Management into Data Management: A System Overview
This paper introduces the CondorJ2 cluster management system. Traditionally, cluster management systems such as Condor employ a process-oriented approach with little or no use of modern database system technology. In contrast, CondorJ2 employs a data-centric, 3-tier web-application architecture for all system functions (e.g., job submission, monitoring and scheduling; node configuration, monitoring and management, etc.) except for job execution. Employing a data-oriented approach allows the core challenge (i.e., managing and coordinating a large set of distributed computing resources) to be transformed from a relatively low-level systems problem into a more abstract, higher-level data management problem. Preliminary results suggest that CondorJ2's use of standard 3-tier software represents a significant step forward to the design and implementation of large clusters (1,000 to 10,000 nodes).
- 1 August 2003
- Published by Springer Nature in The VLDB Journal
- Vol. 12 (2) , 89-101
A case for fractured mirrors
- 1 August 2003
- Published by Springer Nature in The VLDB Journal
- Vol. 12 (2) , 89-101
A case for fractured mirrors
- 1 February 1997
- Published by Springer Nature in The VLDB Journal
- Vol. 6 (1) , 53-72
Data placement in shared-nothing parallel database systems
- 1 February 1997
- Published by Springer Nature in The VLDB Journal
- Vol. 6 (1) , 53-72
Data placement in shared-nothing parallel database systems
- 1 October 1995
- Published by Springer Nature in The VLDB Journal
- Vol. 4 (4) , 629-673
QuickStore: A high performance mapped object store
- 1 October 1995
- Published by Springer Nature in The VLDB Journal
- Vol. 4 (4) , 629-673
QuickStore: A high performance mapped object store