Businesses “have long recognized the value of responding faster—to their customers, and to their operations—but their applications lacked a database that could provide context, awareness, and knowledge fast enough to respond in real time—until in-memory databases were invented,” said Peter Vescuso, CMO of VoltDB. “Memory is getting cheaper faster than operational data stores are growing. When traditional database technologies were designed—many more than 30 years ago—the cost of memory was the biggest constraint. That’s no longer true. Today, many business datasets fit in-memory, so it makes sense—economically and operationally—to use in-memory databases.”
Oracle’s in-memory technology positions the vendor and its products squarely in the middle of the big data and real-time analytics surge.
Fast in-memory databases “are accelerating the speed of business. Most companies simply can’t afford to wait an hour or a day to pull analytics from a data lake; they need to be able to make customer interaction decisions in the moment,” said Vescuso.
Often, relational databases don’t even get considered at all for big data projects. “For the last few years, for most of the requirements large firms have been having with leading-edge projects, they have been using Hadoop, or landing in staging areas,” says Abramson. “They’re reducing their need for relational databases and looking to find more cost-effective solutions for what they’re doing with data.”
While non-relational or NoSQL databases have proliferated for new projects, mixed relational and NoSQL database environments have become the norm within established environments or applications. The result is highly mixed environments across the board. “What I’ve been seeing is people are still using traditional relational databases for a lot of analysis, but NoSQL for the newer stuff,” said Abramson. “If it’s an experimental-type approach, they’re going to implement those with NoSQL. The newer companies are going straight to NoSQL.” In his consulting work, Abramson is seeing companies looking at putting NoSQL up front, with relational databases in the cloud. “That is just so different from anything we would have done 10 years ago,” he said.
While at first blush this would appear to be threatening to Oracle’s business, the vendor’s offerings may play well in these mixed environments. “It all depends on what kind of workload you’re looking at,” said Anand. It all complements each other—from Hadoop to NoSQL, document databases to relational databases, he observed. “There is a great deal of SQL code across enterprises. Oracle has an integration stack, they have data knowledge, they have hardware, and they have security. So if you’re looking at designing and building some sort of a high-throughput, low-latency architecture, the combination of elements that Oracle is providing is a game changer. The Oracle ecosystem provides a lot of depth when it comes to building a true enterprise solution.”
Abramson agrees and observed that “if you want to get high-performance, almost relational-type performance, in an open-type environment, then NoSQL is your solution.” Oracle embraces it all, Abramson said. “They’ve always been a one-stop shop for any of the technologies. Whether it’s big data, structured data or unstructured data, they’ve already provided that. With their acquisitions, they’re buying new tools, new features, and new capabilities. They continue to compete against everybody, and they also compete against themselves."