Non-Relational Databases

November 30, 2018

2 minute read

Of the two major categories of databases, I plan on learning a little bit about non-relational databases first. Before I get into a specific one though, here's a table of the different types of non-relational databases:

Type Example Utilized For
Document Mongo Groups of key-value pairs, able to be nested to depths, fluid schema means flexibility, fast to build on
Key-value Redis Simple key-value stores, scaling well, but does not support complex queries, so suited for simple data
Column Cassandra Designed for retrieval of large amounts of a particular column (attribute), benefiting analytic applications
Graph Neo4J Used for interconnected data and strong at traversing connections between nodes

Document

Document-based databases have fluids schemas, which means individual entries in the database can have differently shaped data. This is really powerful, because it means you can prototype an application rapidly (with data persistence) without needing to design an entire SQL database. When you build a proof of concept or prototype, you know the least about your problem space, so you're probably going to throw away huge swathes of your work, as you learn more about the project. You're building something to learn, so you can iterate into a better version later.

A document-based database like MongoDB is perfect for that. That's why I'm learning it first.

Key-value

Key-value databases are essentially a data dictionary in database format. They can be deployed in cases where you have a simple data structure but need speed and efficiency. The most common use case for this type of database is caching.

Column

I know column-based databases are used in big data and analytics, but I don't really understand the advantages they represent. Frankly, I understand this type the least. Future me will dig into this later. emoji-bowtie

Graph

Graph databases provide a permanent way to store the (wait for it!) graph type of data structure. I know... you're as surprised as I am. Seriously though, this type of database excels at storing things like maps of roads emoji-globe_with_meridians. Future me will investigate further. I've already got a sample project in mind.

Conclusion

Each of these types should be used for different purposes, but the most common all around utility type is a document-based database, because it has a fluid schema. The most popular document database is MongoDB. I am learning it first, because the dynamic nature and API available as part of a document-based database makes it easy to quickly prototype systems and applications. As part of my learning process, I want to build as many things as possible, as quick as I can, and MongoDB enables me to do that.

Checkout my next post, Mongo like Candy, for a deeper dive. I warned you about the throwbacks!