A Framework for Executing Complex Querying for Relational and NoSQL Databases (CQNS)

An IMRAD digest of the CQNS framework for unified complex querying across SQL and NoSQL data stores

Eman A. Khashan; Ali I. el Desouky; S. Elghamrawy· European Journal of Electrical Engineering and Computer Science· 2020· DOI 10.24018/EJECE.2020.4.5.195

The core problem

The increasing volume of data on the web poses major confrontations. The amount of stored data and the number of query data sources have become essential features for large-scale data systems. A wide range of platforms is used to handle the NoSQL database model, including Spark, H2O, and Hadoop HDFS/MapReduce, which are suitable for controlling and managing big data volumes. Developers of different applications impose difficult tasks on data stores by interacting with mixed data models through different APIs and queries. This paper proposes a complex SQL Query and NoSQL (CQNS) framework that acts as an interpreter, sending complex queries received from any data store to its corresponding executable engine. The proposed framework supports application queries and database transformation simultaneously, which in turn speeds up the process. Moreover, CQNS handles many NoSQL databases such as MongoDB and Cassandra. The paper provides a Spark framework capable of handling both SQL and NoSQL databases. It also examines the importance of MongoDB block sharding and composition, while the Cassandra database deals with two types of sections: vertex and edge portioning. Four scenario criteria

Innovation

The proposed CQNS framework was evaluated using four scenario criteria datasets to query various NoSQL databases. The evaluation focused on optimization performance and query execution timing. The results show that among the comparative systems, CQNS achieves optimum latency and productivity in less time. The framework successfully handled complex queries across MongoDB and Cassandra, demonstrating its ability to support application queries and database transformation simultaneously. The experiments also highlighted the importance of MongoDB block sharding and composition, as well as Cassandra's vertex and edge portioning, in achieving efficient query execution. The comparative analysis indicates that CQNS outperforms existing systems in terms of both latency and productivity, validating its effectiveness as a unified querying framework for heterogeneous data stores.

The increasing volume of data on the web poses major confrontations. The amount of stored data and the number of query data sources have become essential features for large-scale data systems. A wide range of platforms is used to handle the NoSQL database model, including Spark, H2O, and Hadoop HDFS/MapReduce, which are suitable for controlling and managing big data volumes. Developers of different applications impose difficult tasks on data stores by interacting with mixed data models through different APIs and queries. This paper proposes a complex SQL Query and NoSQL (CQNS) framework that acts as an interpreter, sending complex queries received from any data store to its corresponding executable engine. The proposed framework supports application queries and database transformation simultaneously, which in turn speeds up the process. Moreover, CQNS handles many NoSQL databases such as MongoDB and Cassandra. The paper provides a Spark framework capable of handling both SQL and NoSQL databases. It also examines the importance of MongoDB block sharding and composition, while the Cassandra database deals with two types of sections: vertex and edge portioning. Four scenario criteria datasets are used to evaluate the proposed CQNS for querying various NoSQL databases in terms of optimization performance and query execution timing. The results show that among the comparative systems, CQNS achieves optimum latency and productivity in less time.

The CQNS framework is designed as an interpreter layer that receives complex queries from any data store and dispatches them to the appropriate executable engine. The architecture supports simultaneous application querying and database transformation, which accelerates the overall process. The framework is built on a Spark framework that can handle both SQL and NoSQL databases. For NoSQL, it specifically addresses MongoDB and Cassandra. MongoDB is examined in terms of block sharding and composition, while Cassandra is analyzed with respect to two types of sections: vertex and edge portioning. The evaluation uses four scenario criteria datasets to query various NoSQL databases, measuring optimization performance and query execution timing. The framework's core mechanism can be represented as a query routing function:

Why it matters

The CQNS framework addresses a critical challenge in modern data management: the need to query across relational and NoSQL databases without sacrificing performance. By acting as an interpreter that routes complex queries to the appropriate engine, CQNS eliminates the overhead of manual query translation and data store-specific APIs. The simultaneous support for application queries and database transformation is a key innovation that speeds up the process. The evaluation results confirm that CQNS achieves optimal latency and productivity, which is particularly important for big data applications where timely insights are crucial. The framework's ability to handle MongoDB and Cassandra, along with its consideration of sharding and portioning strategies, makes it a versatile solution for heterogeneous data environments. However, the paper does not provide detailed quantitative metrics beyond the comparative claim of optimal latency and productivity. Future work could explore additional NoSQL databases and more complex query types. Overall, CQNS represents a significant step toward unified querying in polyglot persistence architectures.

Who should read this

CS practitioners and researchers

Opening member content…