Skip to main content

Big data means the reign of the relational database is over

ITProPortal spoke with Matt Asay of MongoDB about how the 30-year reign of the relational database is over, and giants like Oracle suddenly risk being left behind.

"Relational databases have been around since the seventies," Matt told us, "and they've survived for so long because they do one job well. They think in rows and columns, and because of that they can do some very rigid tasks very quickly, and process structured data well. But they're becoming too rigid for the new kinds of data being produced by the world of big data."

"A company might want to add just a single field to their database, like nicknames for example, or Twitter handles. Now that's 15 minutes' work for a developer to write that code, but it can take upwards of three months just for the database to shift all of its data one column along, and update it across all fields."

"Add to that the fact that some companies routinely have as many as 70 different databases, or even more – and they can be separated geographically. In the tech world agility is so important, and three months is such a long time, that these legacy databases are actually costing companies a lot."

That's where the NoSQL movement comes into its own, Matt told us.

"I think when we're talking about big data, actually the 'volume' in big data is the least interesting. If you look at companies like Google, for instance, they're dealing with relatively small datasets of say 20TB. That's not really what you'd call true 'big data'."

"It's the messiness of the data that causes us problems, and that problem isn't going to go away. It's only going to get worse. Today, unstructured data is growing twice as fast as structured data, and that's why the databases that are part of the NoSQL movement are becoming the right tools for an increasing number of jobs."

"How do you categorise a tweet, for instance, in a database?" Matt asked us, "or all the different kinds of sensor data that's streaming in from all kinds of devices that were never connected in the past? If you think about Google Maps, it might seem like an easy matter to log longitude and latitude and so on, but in fact we're talking about a device that tracks its owner as they move around, and picks up all kinds of other sensor data at the same time."

When asked if the established database giants like Oracle would one day wake up and find themselves behind the curve, Matt said: "It's already happening. Today when people are building applications, the traditional relational databases just aren't getting chosen. What'll probably happen is that Oracle and the others will simply buy someone, once they realise the way things are going. I don't see Oracle, for instance, going out of business. They'll acquire a smaller company. But it won't be us."

So how can someone get started building their projects in MongoDB?

"Just download it. It's free and open source, so give it a go."

That's not to say you can just wade in with no experience, though.

"If someone wants to start out on MongoDB and they have no database experience, I'd probably tell them to get a bit more prepared, but one of the main things developers love about MongoDB is how intuitive it is, and how easy it is to use just out of the box. If you've worked with databases before, you're going to find MongoDB pretty easy to use."

"Basically I'd say choose the right moment. If your boss is coming to you and breathing down your neck, demanding that you deliver a project to a tight deadline – well, now may not be the time to learn MongoDB."

"But basically just download it, and try it out. That's the beauty of open source for me – that's what open source means."

Matt Asay is VP of marketing and business development at MongoDB.