A Logical Database Design Methodology for MongoDB NoSQL Databases

This paper presents a logical database design methodology for a MongoDB NoSQL database. Given a query, the design methodology is able to assist database designers to determine the best set of configurations of data, also known elsewhere as scheme trees, in the database such that the retrieval time of the query can be minimal or reduced. The design methodology first models an application of interest with a conceptual model. Based on our previous researches, the design methodology then generates from the conceptual model as few scheme trees as possible, which will eventually be implemented as MongoDB's collections in the database. To illustrate the design methodology, the COVID-19 data set was downloaded as an example application. The design methodology first conceptualized the data set with an Entity-Relationship model. Multiples queries were then devised to access various parts of the date set, whose executions required retrievals of the attribute values of all or some of the entity types and/or the relationship in the ER model. The design methodology then generated the best sets of scheme trees for the queries.

I. Introduction

NoSQL databases, a next-generation databases that are greatly different from traditional SQL databases, have been specifically developed for the present Big Data era [5], [6]. NoSQL databases offer many benefits that traditional SQL databases cannot [11]. Those that are particularly relevant to this paper are listed as follows: (1)

NoSQL databases are often better suited to storing and modeling structured, semi-structured, and unstructured data in one database. NoSQL databases often store data in a form that is similar to the objects used in applications, reducing the need for translation from the form the data is stored into the form the data takes in the code.

(2)

NoSQL databases were created to handle big data as part of their fundamental architecture. Additional engineering is not required as it is when SQL databases are used to handle web-scale applications. The path to data scalability is straightforward and well understood.

(3)

NoSQL databases are often based on a scale-out strategy, which makes scaling to large data volumes much cheaper than when using the scale-up approach the SQL databases take.