UniqueNOSD: a novel framework for NoSQL over SQL databases

Area of Science:

• Database Systems and Management
• Distributed Computing
• Data Storage and Retrieval

Background:

• Large corporations predominantly use relational databases for core systems and non-relational (NoSQL) databases for non-core systems, prioritizing availability and scalability over consistency.
• NoSQL systems, based on the CAP theorem, often trade consistency for availability and partition tolerance, limiting their use due to lack of robust query engines.
• Existing 'SQL over NoSQL' solutions introduce query layers but suffer from data redundancy and consistency issues, and are not relationally complete.

Purpose of the Study:

• To present a 'Unique NoSQL over SQL Database' (UniqueNOSD) system, an inverse approach to existing methods, enabling full NoSQL functionality without an extra SQL query layer.
• To propose a NoSQL over SQL Block as a Value data storage strategy for document-based NoSQL databases, aiming to eliminate data redundancy and inconsistency.
• To demonstrate improved data storage, retrieval, query execution, consistency, and scalability in large database systems.

Main Methods:

• Introduced a 'NoSQL over SQL Block as a Value' (B as V) data storage strategy, representing relations as [Formula: see text] (a tuple (K, B)) instead of the traditional [Formula: see text] model.
• Defined relations as [Formula: see text] with a key attribute K and a set of n relations (blocks) B, where each block [Formula: see text] contains its own attributes and is related via foreign keys.
• Conducted experiments using benchmark systems of 'SQL over NoSQL', relational databases, and real-life datasets to evaluate the UniqueNOSD system.

Main Results:

• The UniqueNOSD system demonstrated superior performance compared to existing relational databases and 'SQL over NoSQL' systems.
• The proposed system ensures data consistency, scalability, and efficient query execution.
• Significant improvements were observed in data storage and retrieval, without data loss, enhancing overall NoSQL database performance.

Conclusions:

• The Unique NoSQL over SQL Database (UniqueNOSD) system effectively addresses the limitations of traditional NoSQL and 'SQL over NoSQL' approaches.
• The 'NoSQL over SQL Block as a Value' strategy provides a novel solution for consistent and scalable data management in NoSQL databases.
• The findings suggest that UniqueNOSD is a viable and performant alternative for large-scale database systems requiring both NoSQL flexibility and relational data integrity.