| Abstract |
|
This paper describes dynamic load balancing mechanisms for a massively parallel database programming language. It focuses on the revisions made against the previous method devised by the authors. Instead of accurate information on object allocation, the proposed approach utilizes approximate information to determine the relocation candidate objects and the destination. This significantly saves the overhead for collecting load information at runtime. The simulation tests reported in the paper confirm the performance improvement achieved by the proposed approach.
|
 Additional Information
|
Index Terms- parallel processing, database programming language, load balancing, object-oriented database
Citation:
Hirofumi Amano, Ken'ichiro Kimura, Akifumi Makinouchi,
"Improving Dynamic Load Balancing Mechanisms for a Massively Parallel Database Programming Language,"
dante,
p. 431,
1999 International Symposium on Database Applications in Non-Traditional Environments (DANTE'99),
1999
|
