Skip to main content

Do I need NoSQL, SQL or both?

TRP: What problems can arise when only SQL or NoSQL is used?

RH: When you limit the deployment to NoSQL, you are missing out on valuable reporting tools and other business intelligence and analytics functions offered by SQL. When the deployment is limited to SQL, you are missing out on the rapid processing and analysing of huge amounts of unstructured files and data sets offered by NoSQL.

TRP: In what instances is integration not the right way to go?

RH: If you only require the reporting tools offered by SQL or the immediate consistency from NoSQL, avoiding integration is a fine fit.

TRP: What database qualities should one look for when dealing with mission-critical applications?

RH: Mission-critical or business-critical IT – no matter how you say it, it's critical to the organisation and failure is not an option.

For example, major financial institutions cannot afford for applications within their system to fail or even have any amount of down time. It could mean the difference between losing millions of dollars or even threatening the organisation's survival, or ensuring smooth operations.

Case in point, some large enterprises that must manage online transaction processing using NoSQL for unstructured data can run into challenges with Open Source NoSQL databases like MongoDB and Cassandra, which do not support immediate consistency, opting instead for eventually consistent data, which can result in data that is out of sync and inconsistent.

It's critical to understand the limitations of each database option (whether a SQL or NoSQL database) before committing to one version over another. And when it comes to dealing with mission-critical information, scalability, reliability and performance rule the day.

It's also important to look at the company behind the database. Do they have a proven track record with mission critical systems? Are they accustomed to supporting customers in your market sector? Is their support team ready to help solve issues in a timely manner? Is the database being kept current as new APIs and Frameworks come online?

TRP: Is there a specific industry that would benefit most from the integration of multiple database types? If so, which and why?

RH: Just about any industry that has a need to store data, and allow end users to quickly and easily access the data can benefit from the integration of multiple database types, such as financial, transportation, manufacturing, health care, legal, point of sale, to name only a few.

For example, many companies in the financial market handle massive volumes of transactions per day, and terabyte and even petabyte-sized data stores, where the additional processing overhead from a SQL database can significantly increase the hardware requirements necessary to manage the high volume of database transactions per second.

When it comes to handling transactions, immediate consistency – not eventual consistency – is key. Every transaction must be accurately and consistently applied, which means any NoSQL database used for a mission critical financial application should be fully ACID-compliant.

At the same time, the transaction records provide a rich source of analysis on business and market trends. By integrating NoSQL + SQL functionality on the same data store, the financial services firm can efficiently use the right database tool for the right application. Leveraging NoSQL performance for the front-end data capture, and SQL for the back-end reporting and analysis.

TRP: When still running on legacy databases, what options do financial companies have when trying to work with siloed data?

RH: As mentioned earlier, a lot of important data is siloed in legacy applications. Many companies have gone the popular route and completely rewritten their databases, but as expected - that can be extremely time consuming and very costly.