A Design Science Approach to Deletion in Transactional Processing Relational Databases.

摘要

Organizations are wasting money and sacrificing efficiency by buying additional hardware to store their data when what they should be doing is removing unneeded data. However, a problem is that deletion is not often used to reduce data growth in busy transactional processing relational databases (TPRDs) because it greatly increases locking and system load, or requires that the system be taken offline for maintenance operations. This research aimed to provide a way to remove data without having to take the system offline or suffer ongoing locks or increased load.;This study uses the Design Science Research (DSR) communication schema proposed by S. Gregor and A. Hevner in 2013 to develop two artifacts: an improved method for performing online maintenance and analysis activities on TPRDs, and an instantiation of the method within Oracle for deleting data, which can be adapted for use in other relational databases. Some offline methods were evaluated in this research, but were rejected since they did not meet the basic criterion of a zero system downtime. The instantiated artifact was evaluated by comparing it with the SQL "delete" command, which is the standard method for deletion within TPRDs without taking the system offline and the new software artifact the Data Retirement Tool (DRT) proved to be both faster and put significantly less load on the system.;One of the strengths of Design Science Research (DSR) is its emphasis on relevance in the sense that it provides a contribution to knowledge in academia while at the same time solving a real world problem in information systems. The benefit for academia lies in the generalizability of the first artifact, namely the ability to apply the method to a broader set of problems than is solved in the instantiated artifact, specifically that of improved data removal. The benefit for industry lies in the second artifact developed in this study, as the instantiated artifact provides a way to greatly reduce database size without the negative consequences that typically are associated with SQL deletion operations.