download0 view162
twitter facebook

공공누리This item is licensed Korea Open Government License

Title
An empirical study of transaction throughput thrashing across multiple relational DBMSes
Author(s)
서영균Richard T. SnodgrassSabah Currim
Publication Year
2017-06-01
Abstract
Modern DBMSes are designed to support many transactions running simultaneously. DBMS thrashing is indicated by the existence of a sharp drop in transaction throughput. Thrashing behavior in DBMSes is a serious concern to database administrators (DBAs) as well as to DBMS implementers. From an engineering perspective, therefore, it is of critical importance to understand the causal factors of DBMS thrashing. However, understanding the origin of thrashing in modern DBMSes is challenging, due to many factors that may interact with each other. This article aims to better understand the thrashing phenomenon across multiple DBMSes. We identify some of the underlying causes of DBMS thrashing. We then propose a novel structural causal model to explicate the relationships between various factors contributing to DBMS thrashing. Our model derives a number of specific hypotheses to be subsequently tested across DBMSes, providing empirical support for this model as well as important engineering implications for improvements in transaction processing.
Keyword
DBMS Thrashing; Transaction; Throughput; Factors; Structural Causal Model
Journal Title
Information Systems
Citation Volume
66
ISSN
0306-4379
Files in This Item:
There are no files associated with this item.
Appears in Collections:
7. KISTI 연구성과 > 학술지 발표논문
URI
https://repository.kisti.re.kr/handle/10580/14710
http://www.ndsl.kr/ndsl/search/detail/article/articleSearchResultDetail.do?cn=NART77721969
Export
RIS (EndNote)
XLS (Excel)
XML

Browse