Revenge of the Relational Database
There's been a lot of buzz lately about the rise of various types of database engines, collectively referred to as NoSQL databases. Regardless of whether it's a Hadoop, key/value store or document database, they all support SQL queries in one form or another. That leads to one question: Are these database engines a passing fad or simply a technology that has narrow appeal?
As relational databases evolve yet again, organizations will soon be able to layer a document format on top of a relational engine in much the same way columnar database engines are today. At the same time, the amount of data that can be stored in a relational database continues to expand. Given those two developments, the relational database that most organizations already have will soon receive a series of upgrades, which, in many cases, might make the need for additional database engines moot.
There will still be a need for Hadoop to store massive amounts of data, but the idea that Hadoop will replace relational databases as the foundation of a data warehouse application might be a little far-fetched. There also will be times when a dedicated document database makes sense. However, in the longer term, the number of use cases that justify a separate document database in a production environment may be limited.
Bob Muglia, CEO of Snowflake, a provider of a relational database that's delivered as a cloud service, is betting that's exactly what's about to happen. He's is familiar name in the channel because at various points in his career, Muglia ran divisions at Microsoft, including Microsoft Office and Windows Server.
Fresh off raising an additional $100 million in capital, Muglia says notes relational databases have faced technological challenges in the past, only to be eventually extended in a way that absorbed rival emerging technologies. It may take awhile, but that paradigm will soon play out again.
The difference today is that cloud services make it simpler and less expensive to throw compute cycles at various types of applications running on a relational engine. As a result, Muglia contends that relational database models will continue to beat all challengers.
The issue solution providers across the channel will need to wrestle with is the degree to which they think IT organizations want to deploy and support multiple database engines. History would suggest there's some tolerance, but, over time, pressure to reduce costs by consolidating those database engines onto a single platform will almost always carry the day.
Mike Vizard has more than 25 years of experience covering IT issues in a career that includes serving as Director of Strategic Content and Editorial Director for Ziff Davis Enterprise.