SQL or NoSQL?

Databases are used by almost all the applications that we create. If you want your application to store data at some point, then database is a must. But then the question arises which database to use, SQL or NoSQL?

What are SQL databases?

SQL databases are traditional databases that everyone has a picture of in their minds. It has rows and columns. Generally, each row represents an entity and each column of the row represents some attribute of the entity. For example: Consider this table1.

Employee_ID	Employee_Name	Age
001	John Doe	37
002	Jason Fox	29
003	Mary Jane	21

Here, Employee_Name,Age and Employee_ID are attributes of each Employee.

Let us consider a second example. Consider this table2.

Employee_ID	Position	Salary
001	Tech Lead	700000
002	SDE 1	30000
003	Intern	4000

Now, let's say you want to get all the names of people who are at the position of Tech Lead. Note that the second table does not have the names of the Employees, but only Position. So you'll have to perform JOIN between two tables. Both the tables have one column in common: Employee ID. So the query you need is-

SELECT t1.Employee_Name, t2.Position,

FROM table1 t1

INNER JOIN table2 t2

ON t1.Employee_ID=t2.Employee_ID

WHERE Position="Tech Lead"

There result of this query will be:

Employee_ID	Employee Name	Position
001	John Doe	Tech Lead

Thing to note here is that Employee_ID attribute links both the tables together and establishes a relation between them. So SQL databases are also called RDBMS (Relational Database Management Systems).

What are NoSQL databases?

NoSQL databases are different from SQL databases in a way that they don't store data in a traditional row and column format. But in a key-value format or document-oriented format or graph-based format.

For example: MongoDB, an example of a NoSQL database, stores data in the form of documents in the database.

But before creating a database, MongoDB creates a cluster and within this cluster, it creates multiple databases. Within each database, you have multiple documents (collection) and in each document, lies your data.

Cluster (has) -> Databases (have) -> Collection of Documents (have) -> Data.

Each document has some data that is stored in the form of JSON. An example of JSON would be

{
    "name":"Jhon Doe",
    "Position":"Tech Lead",
    "Age": 37,
    "Salary":"700000",
}

So each person in the document is a JavaScript Object. So the entire table1 if stored in MongoDB would look like this.

{
    "Jhon Doe": {
        "Position": "Tech Lead",
        "Age": 37,
        "Salary": 700000
    },
    "Jason Fox": {
        "Position": "SDE 1",
        "Age": 29,
        "Salary": 30000
    },
    "Mary Jane": {
        "Position": "Intern",
        "Age": 21,
        "Salary": 4000
    }
}

Each employee is a JSON object and each object has some fields associated with it. Now, if you want to query some data in the document, the query would go like:

db.employeedata.find( { "position" : "Tech Lead" } )

Here employeedata is the name of the collection where our data is stored. This query is equivalent to the SQL query above.

The thing to note here is that as data is stored in JSON format, adding a value becomes easy. For example, If you want to add attribute End Date of internship for Mary Jane, so the table becomes:

"Mary Jane": {
    "Position": "Intern",
    "Age": 21,
    "Salary": 4000,
    "End Date":12-07-2020
}

The same goes for deletion. If you delete the Salary field from Mary Jane, the entire database does not get affected. But in case of SQL, if you want to delete Salary of an entity, you have to delete Salary for all the entities.

Differences between SQL and NoSQL

No.	SQL	NoSQL
1	Fixed Schema (Row and Columns)	Flexible Schema (JSON)
2	Useful to make Complex Queries (JOINS)	Complex queries are not possible
3	Follow ACID properties	ACID properties not guaranteed
4	Vertically Scalable (By increasing CPU/RAM)	Horizontally Scalable (By adding more servers to pool)
5	MySQL, MS-SQL, Oracle, PostgreSQL	MongoDB, Cassandra, Redis

Conclusion

If you want to develop an application that will be storing sensitive data or any other application where you think that data should be consistent throughout the database (ACID properties), you should go for SQL database. That is why databases used by banking applications, Air Ticket booking websites, and other online booking sites always use SQL.
If you are developing an application where security is not an issue, or you expect that user is not going to perform complex queries such as JOINS, or maybe just you want to have a flexible schema for database, go for NoSQL database. That is why database used by EA Sports for storing the user preferences and user data, is NoSQL.