Deciding Between SQL and NoSQL for Document-Oriented Databases in Microsoft Azure

Admir Mujkic
6 min readMar 17, 2024


At the heart of the SQL vs NoSQL decision is the structure of your data.

Deciding Between SQL and NoSQL for Document-Oriented Databases in Microsoft Azure

SQL Databases or The Orderly World

Once upon a time, there was a world where everything had its place, and everything was in its place. SQL databases live in such a world. It is as if they were an organized library in which each book is in its place on its own particular shelf. SQL databases expect the data to fit into structures defined beforehand.

They are best suited to data that remains consistent over time, provides for complex queries, and maintains data integrality through relationships and constraints.

Let’s consider real world use-case

Having been involved in the fintech sector, particularly the banking industry, before a decade, I will use this experience as a use case. Such systems provide excellent examples for explaining SQL Databases.

Imagine a large banking institution that provides various services such as savings and checking accounts, credit cards, loans, investment services, etc. The bank’s data includes customer information, account balances and transactions, loan repayment transactions, interest calculation issues, etc.

What will be main requirements?

Data Integrity

All financial transactions must be recorded and executed accurately. Any mistake could lead to significant financial losses or legal and regulatory risks.

Complex Querying

The bank should be able to generate complete reports of account usage and organization-level reporting, in addition to inquiring about customer spending and other operations. This requires the ability to query over millions of transactions in addition to aggregating activities in a customer book, processing offers to customers, etc.

Security and access control

Security is essential, and multiple roles must be involved. For example, a CSA has access to customer profiles but not to the bank’s financial reports.

Now we should ask why SQL databases?

SQL Databases are well-suited to such applications for the following reasons:

Predefined schema

Each form of transaction or account has a structured layout, e.g., loan account contains attributes such as principal amount, interest percentages and repayment terms, etc., different from a savings account.

Transactions and ACID properties

SQL Databases enable developers to utilize property verified transactions that are consistent with the liquidity, sovereignty, separation and strength of operations processes. Financial operations are handled in this safe manner. The functionality allows operations to be passed back to a committed state in the event of a crash, preventing transaction partitions that result in useless state.

Complex Reporting and Analysis

SQL enables banks to easily generate complex analytic statements such as differen reports or information in all my savings accounts for the month or eligibility data offer to the refunds of people who have an extension of 25,000 and the last loan opened more than six months ago.

A bank utilizes an SQL database such as Microsoft SQL Server or Oracle to operate its core banking database. Each customer is saved in a customers table that is connected to various accounts in the same accounts table. The transactions that flow from tables to tables use foreign keys to heal the earnings of respective accounts.

Developers can manipulate and manipulate the response sufficiently to provide the foundation of good execution, data stockpiling and protection of the connected data storage configuration in the SQL database.

Thus, SQL databases are the perfect database for banking and finance because it allows for precisely errorless procedures with all data simultaneously saved in a friendly campaign.

NoSQL Databases

On the other hand, NoSQL databases are designed to be scaled horizontally. They can easily spread across a number of servers, which makes them applicable for cloud environments and large and scattered systems. Additionally, by keeping the related data together, they lower the necessity for complex joins, which is especially useful for read-heavy apps.

Let’s consider real world use-case

Let’s image an e-commerce platform, such as Amazon or eBay, that operates a broad product catalogue and collection of users. It must deal effectively with numerous data varieties such as profiles, product descriptions, procurement history, and suggested action — particularly during periods of active purchasing.

Here’s where Azure Cosmos DB, a NoSQL database service from Microsoft, can help a lot.

How Can E-Commerce Overcome Scalability Challenges and Manage Varied Data Effectively?

An e-commerce platform has to handle a growing number of products and user-related data. Products come with numerous attributes, and individual user profiles contain distinct information — from several shipping addresses to payment methods. When the sale or holiday season comes, traffic on the website booms, and the database must support dynamic scaling to maintain smooth operation.

Azure Cosmos DB for Global Distribution and Flexibility

Azure Cosmos DB is the one of the best choice for use cases involving global distribution and horizontal scaling. For instance, a global e-commerce platform that has a diverse customer base can utilize Cosmos DB to replicate their data in multiple regions.

This way, the e-commerce platform will guarantee limited latency in accessing real-time data regardless of geographic location of the user. As load increases, more regions can be added to Cosmos DB without any downtime, to ensure that the platform remains available and responsive to the users.

The e-commerce platform schema-less feature allows it to add new product types or extend root field values without making any changes to the database. This makes sure that the e-commerce platform adds new features or new categories without any interruption.

Solution as Azure Cosmos DB

As a solution using Azure Cosmos DB for an e-commerce platform can distribute its data globally, ensuring users have fast access to the latest product listings and their own data, such as wish lists and cart items, from anywhere in the world.

The ability to scale horizontally allows the platform to handle increased loads during sales events efficiently, ensuring consistent performance and user satisfaction.

Moreover, the multi-model capability and schema-less design offered by Azure Cosmos DB enable the e-commerce platform to rapidly adapt to market trends and user demands, introducing new features and product types without extensive backend reworks.

Azure Cosmos DB provides a robust, scalable, and flexible database solution for e-commerce platforms, supporting them in delivering a seamless and responsive user experience across the globe.

Pros and Cons SQL V.S. NoSQL

The Best of Both Worlds or Hybrid Solutions

The distinction between SQL and NoSQL is not always black and white. Many organizations opt for a hybrid approach, using SQL databases for their core structured data needs while leveraging NoSQL for more flexible, scalable components. This approach allows businesses to enjoy the benefits of both worlds, tailoring their database strategy to fit the unique needs of each application component.

Final Words or Better Said a Thoughtful Choice

Ultimately, the choice between SQL and NoSQL databases is determined by your project’s requirements, data type, expected scalability, and query complexity.

SQL databases are reliable and have a powerful query language for structuring data. In another hand, NoSQL databases are flexible and can scale horizontally, making them appealing for unstructured data and distributed systems.

An understanding of both types advantages and disadvantages will allow you to determine which one will meet your application demands and ensure its development.



Admir Mujkic

Admir combined engineering expertise with business acumen to make a positive impact & share knowledge. Dedicated to educating the next generation of leaders.