Concurrent access to O-trees
Technical Report
·
OSTI ID:5650561
With the advent of parallel computing, issues of concurrent access to, and manipulation of, database index structures have become increasingly vital. In this thesis, we introduce a highly optimistic method for concurrent access to indices that involves no locking for searching, insertion, splitting, or simple deletion and minimal locking for merging. The method uses node version numbers to control the writing of nodes and depends heavily upon retry techniques. The index structure used is the O-tree, developed by Ratko Orlandic. We perform an analysis of the algorithm and the number of messages required to perform insertions using the locking algorithm of Lehman Yao and the optimistic approach we present. The analysis shows a marked decrease in message traffic using our retry techniques for a reasonable number of processes inserting entries into the same tree.
- Research Organization:
- Virginia Univ., Charlottesville, VA (USA)
- Sponsoring Organization:
- DOE; USDOE, Washington, DC (USA)
- DOE Contract Number:
- FG05-88ER25063
- OSTI ID:
- 5650561
- Report Number(s):
- DOE/ER/25063-T12; ON: DE91014348
- Country of Publication:
- United States
- Language:
- English
Similar Records
On the performance of concurrent tree algorithms
Fast fault-tolerant concurrent access to shared objects
Development of a concurrent tree search program
Journal Article
·
Wed Feb 28 23:00:00 EST 1990
· Journal of Parallel and Distributed Computing; (USA)
·
OSTI ID:6779539
Fast fault-tolerant concurrent access to shared objects
Conference
·
Mon Dec 30 23:00:00 EST 1996
·
OSTI ID:457685
Development of a concurrent tree search program
Technical Report
·
Fri Oct 01 00:00:00 EDT 1982
·
OSTI ID:7091838