Replicating time-invariant data in a distributed database environment
Abstract
This dissertation presents Time-Invariant Fragmentation, a novel data replication approach that takes into account the time sensitivity property of data. A time-invariant fragment (TIF) is a portion of a database the values of which do not change during a specified time interval. Next, it presents the algorithms for determining time intervals and creating TIFs for the time intervals and processing queries in the TIF environment. The benefits of the TIF approach are explored through two simulations. The first simulation compares the TIF approach with three other approaches, namely, full-replication, non-replication, and materialized view, in terms of data transmission cost. Results indicate that TIF is more effective when the percentage of modification queries is high, the network size is large, and the number of queries is high. The second simulation further considers the data storage cost and overhead cost for creating TIFs and processing queries in the TIF environment. It generates the database and query history and applies the TIF algorithms to determine time intervals and the TIFs. It also applies time intervals to materialized view approach. Results indicate the performance of both TIF and materialized view has improved. The simulation also provides a mechanism for choosing the best data replication portfolio.
Degree
Ph.D.
Advisors
Chaturvedi, Purdue University.
Subject Area
Management|Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.