Home // DBKDA 2012, The Fourth International Conference on Advances in Databases, Knowledge, and Data Applications // View article

Dynamic Stream Allocation with the Discrepancy between Data Access Time and CPU Usage Time

Authors:
Sayaka Akioka
Yoichi Muraoka
Hayato Yamana

Keywords: load balancing, resource management, stream mining

Abstract:
Huge quantities of data arriving in chronological order are one of the most important information resources, and stream mining algorithms are developed especially for the analysis of the fast streams of data. A stream mining algorithm usually refers to the input data only once and never revisits them (read-once-write-once), while the conventional data intensive applications refer to the input data in a write-once-read-many manner. That is, once the stream mining falls behind, the process drops the input data until it catches up with the input data stream. Therefore, the fast execution of the stream mining leads the perfect analysis on all the input data, and it is very critical for the quality of the service. We propose a dynamic resource management for the stream mining in the distributed environment. The resource management utilizes the discrepancy between data access time and CPU usage time inside the stream mining, and speeds up the mining process. We implemented the methodology and proved successfully to process all the input data of such a fast data stream, whereas the serial execution drops more than 90% of the input data.

Pages: 169 to 174

Copyright: Copyright (c) IARIA, 2012

Publication date: February 29, 2012

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-61208-185-4

Location: Saint Gilles, Reunion

Dates: from February 29, 2012 to March 5, 2012