Scalable data access layer over non-relational databases

OData support
Dr. Simon Vilmos
Department of Networked Systems and Services

Nowadays web applications managing great quantities of data (search engines, social sites, mailing and content sharing systems) have made it important to switch the well tried and boiled down relational data model to solutions better suiting their specific needs. The aspects which made the paradigm switch necessary were the increase of scalability, redundancy or partition tolerability. Many database systems have been born in the past few years, aiming to give a solution to these problems sometimes using different approaches. These software systems are referred to as non-relational or NoSQL by the scientific literature.

One of the main strengths of relational database systems is the existence of the SQL language. There is no such common language or abstraction for non-relational systems. This is partly because of the variety of these systems, partly because of the approach fundamentally different approach from SQL, and partly because of the novelty of these technologies.

In the thesis, I analyse the possibilities to offer a general abstraction over a set of non-relational systems. I implement a software layer, on whose standard interface applications which need these kind of databases can be built, or the concretely used non-relational database systems can be exchanged below it. The layer to be implemented can be compared to the Object Relational Mapping [15] type design pattern, known from the relational world, which maps an object hierarchy to a data model.


Please sign in to download the files of this thesis.