The dilemma of choosing schema or schema-less databases. This post aims to simplify this key design decision.
The schema based databases has stood the test of time, in fact Relational Database Management Systems (RDBMS), have been a pioneer in this area.
Since the 1980s, and to the time of of this post, this type of database has been harnessed for:
And many other applications.
However, schema based databases are not perfect, in fact they did encounter issues during there time, which we continue to battle today. These issues include, the object-relational impedance mismatch between relational database and the object-oriented applications and many others that I will expand on later.
Additionally, there were attempts to address object-relational impedance mismatch issue with schema based databases, with the introduction of object database managements systems, which alleviated this issue particularly. An example of this is Object Relational Mapping (ORM) which in Java this is called the Java Persistence API, that is used for many enterprise Java applications.
Whilst this had success, the expanse of technology and challenges continued, such as horizontal scaling of node clusters. This resulted in the birth of NoSQL (schema-less databases).
The rise of NoSQL was during the 21st century and this gained popularity with the NoSQL hashtag, generally this was for those whom wanted to discuss new types of databases.
Furthermore, it is important to note that NoSQL databases are generally implemented with partial schemas, that are enforced by the Database Management System (DBMS) on a write basis, whilst with the integration of the application acting as an externalised schema on reads.
Although the name NoSQL does not have a direct connotations to what it is, it does however require it to be schema-less and typically the data store to be accessible using key value pairs.
My final footnote, to warn this topic is highly variable when comparing schema and schema-less database systems, due to how different technologies have implemented this. Although, this pros and cons list will give you good gauge of what to consider when making the database design decision.
So just to round off, we can say that schema based databases are generally less flexible, but have added built in security characteristics. It is also noted, that the lack of flexibility is not entirely a disadvantage, this maybe more of an advantage due to the application use cases / requirements. Whilst in comparison, NoSQL is more preferable for those seeking more flexibility with the database and increased ability to handle larger volumes of data, with the added advantage of getting it set up much faster, although this needs to compared to the application requirements before, as one advantage can be a disadvantage to another with also the technology implementations. Which in turn, makes this pros and cons list more variable but a good source of understanding the differences.
