Technology has made leaps and bounds since the brick sized mobile phones a handful of us carried around in the 70s. Just walk down the street and count how many people are holding a device in their hands.
Front-end technology innovation is immensely visual and prolific, but are people considering back-end innovations? Databases and storage systems are hidden from the masses and as such, businesses are allowing systems to wallow in the past, languishing in architectures that were relevant for 1960s and 1970s technologies, but not for today.
Look at your legacy database architecture. These databases were top-notch four decades ago when there was nice flat data that could fit into rows and columns. While there are a few instances today in which these relational databases are still valuable, mostly they simply hinder processes and cost money.
All businesses are aware that critical data affects bottom line revenue or customer experience but it has changed formats since its beginnings. Today’s data can be in the form of items to add to an online shopping cart, tweets, or real-time stats for a mobile game, to name a few. Like data, customer expectations have also changed: the Internet and complementary technologies insist that big and critical data be up-to-date and immediately available.
This means that downtime equals a steady decline or even death of a business. In order to compete in today’s highly evolved market, businesses need to ensure their system provides 24×7 availability, low-latency regardless of customer location, and scalability to handle inconceivable growth, even within a moment’s notice.
The old relational databases were built to manage specific data types, and are not agile enough to handle these new demands. Businesses now have a new option with the emergence of distributed systems that better manage data, and prevent datacentre downtime and failure.
To be sure, relational databases have their uses as shown by Oracle and IBM holding their ground. For example, big banks in particular deal with flat, structured data that needs to be stored and retrieved for decades in relational databases, particularly in today’s litigious society.
However, most data today does not require this structured data model but simply some consideration as to how to model it from the get-go. Businesses must evaluate the type of data they work with to decide the level of availability needed on a regular basis.
For use cases that require data to always be read/write available, a traditional database simply isn’t the right option. These use cases can be as simple as a website requiring 24×7 availability, or as complex as retrieving life-saving patient data in a new universal health care system.
Another modern database concept for businesses to consider is horizontal scaling rather than vertical. With a relational database, vertical scaling requires buying another costly database and then manually sharing data between machines. This process is operationally and financially intensive, and may still result in hot spots. This is part of why wildly successful companies like Amazon and Twitter run on non-relational databases.
Distributed systems scale horizontally, as all nodes are equal, with no concept of a “master” node. These systems run on commodity, interoperable hardware and individual nodes can simply and quickly be added or removed based on traffic spikes — eliminating unnecessary hardware and operations costs.
So why are Oracle and IBM still thriving? Traditional databases have been around for decades, and operations teams are very familiar with them. Businesses have also convinced their businesses to invest heavily in them. As such, some businesses tend to hold onto their relational systems, rather than be proactive and search for alternative options, even if it costs the company millions.
However, for companies to truly innovate and remain competitive, businesses need to learn about distributed systems and start integrating them into their infrastructures today. It may be just starting small with an internal tool, and then expanding from there. For most use cases, they will discover that distributed systems are cheaper and easier to manage than existing legacy systems.
It’s time to throw out the old overpriced and oversold and bring your architecture into the brave new world.