Cloud Computing Comparison: PaaS Providers

Relational databases have dominated the storage scene for the last 40 years, but they don't hold unstructured or semi-structured data very well. That kind of information--including audio and files, blog posts, data from sensors on mobile devices, and social media streams--is becoming increasingly common, and organizations must find innovative ways to manage it.

The shortcomings of the relational database--at least from a big data perspective--have led to the emergence of a variety of competing database technologies such as NoSQL, a schema-less alternative that offers an enterprise a lot more flexibility in how it stores data.

But relational technology may have a big data future after all. JustOne Database, a tech startup with offices in Connecticut and the United Kingdom, has developed a new relational database that it says overcomes many of the technology's weaknesses. Its NewSQL-based JustOneDB handles billions of new transactions per day, supports analytics as well as fast transactions and queries, and runs on low-cost commodity hardware.

"This, to me, is the most remarkable innovation that I've ever seen in databases in the decades I've been in the business," claimed JustOne Database CEO Kate Mitchell in a phone interview with InformationWeek. A veteran of IBM and Oracle, Mitchell and her two co-founders launched JustOne Database three years ago.

[ Learn why it's tough to keep up with all the unstructured data coming into your systems. See Big Data Means Big Storage Choices. ]

So what makes JustOneDB unique?

"There's one thing that's probably the single biggest differentiator for us," Mitchell said. To wit, JustOneDB is the first "genuine" general-purpose relational database to support both analytics and transactional queries in the same database.

But don't several major industry players--including tech heavyweights with household names like IBM and Oracle--sell relational databases too?

Yes, Mitchell said, but when you use one of their databases for transactions, you don't use the same physical database for analytics. Hence they're not as "general purpose" as they claim.

"You actually do the analytics in a separate database, and go through a process called an 'extract, transform, and load,'" Mitchell explained.

This process involves pulling selective data out of the transaction database and reorienting it. "You store it in a very different way, so that when you're doing queries in your analytics system, your queries are going to perform well," she said.

With JustOneDB, however, you don't need to pull data out and load it into a separate database. You also can do any query at any time without planning in advance. And JustOneDB has a unique storage model that isn't row or column based; rather it stores data by relationship, the company said.

It remains to be seen, however, if JustOneDB can live up to its hype. One major reason is that the company has released only half of the product thus far.

"We decided to build the first piece--the transaction side, get some traction and customers and credibility, and then come along and do the next piece: true analytics," said Mitchell.

The analytics component of JustOneDB will tackle the "really broad kind of analytics that people do today, such as, 'I want to see all the people in this zip code--with this kind of income--who bought products between these days,'" she said.

The transaction portion is available today from several platform-as-service vendors, including Engine Yard, AppHarbor, and the Salesforce.com-owned Heroku.

JustOne Database hasn't announced a release date for the enterprise version of JustOneDB, which will include both the transaction and analytics pieces. About 300 companies currently use the cloud version of the product.

Relational technology's share of the database market will likely decline in the coming years, as organizations adopt alternative products such as NoSQL for big data management. This doesn't mean that the relational database is history, however.

"The need for relational databases to power businesses and entities is not going away," Mitchell said. "For businesses, large government entities, and the healthcare and public sectors, the relational database really runs most companies around the world."

New innovative products may be a better fit for today's enterprise storage than monolithic systems. Also in the new, all-digital Storage Innovation issue of InformationWeek: Compliance in the cloud era. (Free with registration.)