A Quick Introduction to Relational & Non-Relational Databases

In 2017 alone, the world is on track to create more data than has been produced since the dawn of human history.  While it has long held that data processing and storage capabilities increase in accordance with Moore’s Law, we are rapidly approaching finite physical limits on hardware.  This problem has driven engineers to develop software solutions—new database architectures—that have given rise to the variety of data management systems we have available today.

There are two kinds of data management systems: relational databases, and non-relational databases. A relational database is a data management system wherein related data are stored in distinct tables from which they can be accessed or reassembled in many different ways according to the developer-defined relationships between tables. A non-relational database is any database architecture not built around tabular relations.  These types of databases come is a variety of forms, and typically contain large volumes of unstructured or semi-structured data.

The origination of a relational database provides strong insight in the current framework of computing demand. The relational database paradigm was invented by E.F. Codd while he was working at IBM in 1970. Then, in 1974, two other IBM employees named Donald D. Chamberlin and Raymond F. Boyce released SQL or “Standard Query Language” which allowed engineers to efficiently interface with relational databases. By the mid 1970’s, the world was beginning to take notice of the powerful combination of SQL and relational databases and in 1979, a fledgling software company by the name of Software Development Labs released the first commercially available implementation of these technologies.  That company would eventually grow into the multi-billion dollar institution named after its flagship database management system—Oracle. Until recently, Oracle’s efficacy and lack of any relevant competition allowed the software, and by extension, the relational database model, to take root as the standard across industries.

The relational database’s counterpart took high-frequency trading to a new level, and although non-relational and relational databases work best in conjunction, non-relational database initially acted as a strong competitor. Non-relational databases, though first conceived of in the 1960’s, did not become popular until the past decade.  The explosion in data creation facilitated by increased global internet access, the prevalence of social media, and improvements in computing power demanded more flexible data management systems than relational databases could offer. Non-relational databases are able to offer increased scalability, performance, and adaptability relative to their relational counterparts.  However, these benefits often come at the cost of desirable characteristics like data integrity, security, or consistency.

The application of these databases to the financial services sector is powerful. Modern financial institutions tend to employ both technologies in their operations. For operations requiring high security and data integrity such as transaction processing and validating between bank accounts, relational databases are the go-to.  For analysis of data such as high-frequency trading where there is high throughput and every millisecond affects the bottom line, one might opt to use a non-relational database schema. At the end of the day, data management tools are only as powerful as the developers working with them.  Certain tools are better suited for certain projects and doing due diligence to determine beforehand which tool is best can make a significant difference in the long run.

Blog Categories:

Add new comment