What is NoSQL?
The term NoSQL was introduced in 2009 and describes a type of non-relational, distributed database. NoSQL databases can efficiently query and serve substantial amounts of semi-structured and unstructured data. Unlike relational databases, NoSQL databases are schema-agnostic. NoSQL databases don’t define a specific columnar structure, which allows for greater flexibility, less brittle systems, and support for rapidly changing or unstructured data.
In the diagram above, you can see that relational tables have a defined schema. Specific columns and their types represent each table’s structure, and relationships are represented via foreign keys. For example, a single Orders row may have many Order Details rows associated with it. To query this information in its entirety, you must join both tables.
In stark contrast, the NoSQL document below depicts the same Order/Order Details relationship via an internal object array in a single document.
{
"orderId": "873sdfs42981",
"orderDate": 1574161910220,
"firstName": "Zhenis",
"lastName": "Omar",
"address": "123 Broad Street",
"city": "Seattle",
"state": "WA",
"postalCode": "92839",
"orderDetails": [
{
"orderDetailId": "ds23fhjsdk1",
"productId": "sdf2378",
"unitPrice": 7.99,
"quantity": 1
},
{
"orderDetailId": "zct687786es",
"productId": "d87f98z",
"unitPrice": 4.59,
"quantity": 3
}
]
}
Retrieving a single Order and its associated Order Details involves the lookup of a single document.
Benefits of NoSQL
NoSQL databases are open-source, distributed, and non-relational. These characteristics lend themselves to the following benefits:
- Non-relational allows storage of semi-structured and unstructured data
- Schema-less, which allows flexible data models
- Rapidly adaptive to changing requirements
- Horizontally scalable due to the distributed nature
- Highly available due to the distributed nature
If you’re coming from a relational background and still aren’t sure how NoSQL differs, see Understanding the differences between NoSQL and relational databases.
NoSQL options
The NoSQL space offers many options. When choosing a NoSQL database system, consider what kind of data you’ll work with:
- Key-value data
- Column-family data
- Document data
- Graph data
- Schema-less JSON
In this guide, we’ll introduce you to Azure Cosmos DB, Microsoft’s flagship non-relational database that supports each of these data options. You’ll see an application that uses Azure Cosmos DB for NoSQL to work with schema-less JSON.
NoSQL for our scenario
You’ll see the benefits of NoSQL in our Contoso pet store eCommerce example.
While the data may seem relational, there are cases for non-relational stores. New product information may come from vendors in various formats: structured, semi-structured, or unstructured. You may need to scale your operations for large traffic, such as running sales on an eCommerce site. Reporting on this data may be complicated.