Date of Award

11-2017

Document Type

Thesis

Degree Name

Master of Science and Software Engineering (MSSE)

Department

Computer Science

First Advisor

Dr. Mamoon Awad

Second Advisor

Salah Bouktif

Third Advisor

Zaher Al Aghbari

Abstract

Most of the enterprises that are dealing with big data are moving towards using

NoSQL data structures to represent data. Converting existing SQL structures to

NoSQL structure is a very important task where we should guarantee both better

Performance and accurate data. The main objective of this thesis is to highlight the

most suitable NoSQL structure to migrate from relational Database in terms of high

performance in reading data. Different combinations of NoSQL structures have been tested and compared with SQL structure to be able to conclude the best design to use.For SQL structure, we used the MySQL data that is stored in five tables with different types of relationships among them. For NoSQL, we implemented three different MongoDB structures. We considered combinations of different levels of embedding documents and reference relationships between documents.

Our experiments showed that using a mix of one level embedded document with a

reference relationship with another document is the best structure to choose. We have used a database that contains five tables with a variety of relationships many-to-one, and many-to-many. Also the huge amount of data stored in all the structures about 2 millions record/document. The research compares clearly between the performances of retrieving data from different MongDB representation of data and the result shows that in some cases using more than one collection to represent huge data with complex relationships is better than keeping all the data in one document.

Share

COinS