Multi-dimensional access control for distributed column-oriented databases

Vast, E. (2013) Multi-dimensional access control for distributed column-oriented databases. Master's Thesis / Essay, Computing Science.

Preview	Text Vast.E.Thesis.pdf - Published Version Download (695kB) | Preview
	Text AkkoordLazovik.pdf - Other Restricted to Registered users only Download (26kB)

Abstract

Distributed NoSQL databases are increasingly used for data mining. One technique is to enrich analysis results with data of other databases. In environments as the government these databases can contain sensitive data. Most relational databases support fine-grained access control for individual data elements. In distributed databases this leads to significant performance loss when the results and access rights are stored on separate machines. Furthermore, the original rights can apply to different dimensions in the merged dataset. This work proposes multi-dimensional access control for column-oriented databases. It is based on role based access control to secure a merged dataset. The access rights are stored within each database cell for the best performance. A prototype based on Apache HBase compares the cell location with two other locations: a separate table and in each row. The locations are compared quantitative with the Yahoo! Cloud Serving Benchmark on a computer cluster of 14 machines. The results show that the location of the access rights has a significant influence on the performance. When the access rights and the data are stored physically closer, both throughput and disk usage increase. The best location depends on the required access control dimension. For a merged dataset where multiple dimensions have to be checked, storing the rights in each cell has the best performance.

Item Type:	Thesis (Master's Thesis / Essay)
Degree programme:	Computing Science
Thesis type:	Master's Thesis / Essay
Language:	English
Date Deposited:	15 Feb 2018 07:52
Last Modified:	15 Feb 2018 07:52
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/10866

Actions (login required)

View Item