NoSQL stands for Not-only-SQL. These are database systems that are not based on a relational approach. The architecture of NoSQL databases usually aims at scalability and performance, their relevance is especially high for Big Data applications. While relational database systems strive for or ensure high data consistency, consistency is of secondary importance in NoSQL databases. The ACID model (Atomicity, Consistency, Isolation, Durability) does not apply to NoSQL databases, rather the BASE model (Basically Available, Soft State, Eventually Consistent).
Relational databases (e.g. Oracle, MS SQL Server, MS Access, MySQL) contain tables with columns and rows. In NoSQL databases, data is stored according to value pairs, documents or lists; actually, the variety is almost unlimited, here are just two examples, I quote from an article of the online magazine "Big Data Insider": "The Key-Value-Database model uses simple keys and value schemes. Individual values are always assigned to specific keys and can consist of arbitrary character strings or structured characters. The Redis database solution uses the Key-Value database model. In a column oriented database system entries are stored in columns instead of rows. The advantage of this model is that in contrast to line-oriented databases no unnecessary data has to be read and read processes are faster.".