By allowing companies to scale “out” in distributed fashion rather than scaling “up” via additional expensive servers, “it makes it possible to very cheaply put together a large data collection and then see what you’ve got,” Olofson said.
Alternatives and favourites
Among other new RDBMS alternatives are the NoSQL family of offerings, including MongoDB – currently the fourth most popular database management system, according to DB-Engines – and MarkLogic.
“Relational has been a great technology for 30 years, but it was built in a different era with different technological constraints and different market needs,” said Joe Pasqua, MarkLogic’s executive vice president for products.
Big Data is not homogeneous, he said, yet in many traditional technologies, that’s still a fundamental requirement.
“Imagine the only program you had on your laptop was Excel,” Pasqua said. “Imagine you want to keep track of network of friends – or you’re writing a contract. Those don’t fit into rows and columns.”
Combining data sets can be particularly tricky.
“Relational says that before you bring all these data sets together, you have to decide how you’re going to line up all the columns,” he added. “We can take in any format or structure and start using it immediately.”
NoSQL databases don’t use a relational data model, and they typically have no SQL interface. Whereas many NoSQL stores compromise consistency in favor of speed and other factors, MarkLogic pitches its own offering as a more consistency-minded option tailored for enterprises.
There’s considerable growth in store for the NoSQL market, according to Market Research Media, but not everyone thinks it’s the right approach – at least, not in all cases.
NoSQL systems “solved many problems with their scale-out architecture, but they threw out SQL,” said Monte Zweben, Splice Machine’s CEO. That, in turn, poses a problem for existing code.
NewSQL
Splice Machine is an example of a different class of alternatives known as NewSQL – another category expecting strong growth in the years ahead.
“Our philosophy is to keep the SQL but add the scale-out architecture,” Zweben said. “It’s time for something new, but we’re trying to make it so people don’t have to rewrite their stuff.”
Deep Information Sciences has also chosen to stick with SQL, but it takes yet another approach.
The company’s DeepSQL database uses the same application programming interface (API) and relational model as MySQL, meaning that no application changes are required in order to use it. But it addresses data in a different way, using machine learning.
DeepSQL can automatically adapt for physical, virtual or cloud hosts using any workload combination, the company says, thereby eliminating the need for manual database optimisation.
Among the results are greatly increased performance as well as the ability to scale “into the hundreds of billions of rows,” said Chad Jones, the company’s chief strategy officer.
An altogether different approach comes from Algebraix Data, which says it has developed the first truly mathematical foundation for data.
Whereas computer hardware is modelled mathematically before it’s built, that’s not the case with software, said Algebraix CEO Charles Silver.
“Software, and especially data, has never been built on a mathematical foundation,” he said. “Software has largely been a matter of linguistics.”
Following five years of R&D, Algebraix has created what it calls an “algebra of data” that taps mathematical set theory for “a universal language of data,” Silver said.
“The dirty little secret of big data is that data still sits in little silos that don’t mesh with other data,” Silver explained. “We’ve proven it can all be represented mathematically, so it all integrates.”
New platform
Equipped with a platform built on that foundation, Algebraix now offers companies business analytics as a service. Improved performance, capacity and speed are all among the benefits Algebraix promises.
Time will tell which new contenders succeed and which do not, but in the meantime, long-time leaders such as Oracle aren’t exactly standing still.
“Software is a very fashion-conscious industry,” said Andrew Mendelsohn, executive vice president for Oracle Database Server Technologies. “Things often go from popular to unpopular and back to popular again.”
Many of today’s start-ups are “bringing back the same old stuff with a little polish or spin on it,” he said. “It’s a new generation of kids coming out of school and reinventing things.”
SQL is “the only language that lets business analysts ask questions and get answers — they don’t have to be programmers,” Mendelsohn said. “The big market will always be relational.”
As for new types of data, relational database products evolved to support unstructured data back in the 1990s, he said. In 2013, Oracle’s namesake database added support for JSON (JavaScript Object Notation) in version 12c.
Subscribers 0
Fans 0
Followers 0
Followers