In late June, Don Rippert left a nearly 30-year career at Accenture, where he was CTO, to serve as CEO of NoSQL database startup Basho. But why would someone leave a top role at a Fortune 500 company for a position at a 40-person startup in a fairly niche market? According to Rippert, it’s because the work is fun and there’s the potential to be part of an enterprise IT revolution.
One of Rippert’s responsibilities at Accenture was leading the company’s Technology Lab, through which, Rippert said, he was frequently coming across big data and NoSQL. His team’s job was to research, analyze and prototype these new technologies so Accenture would be ready to implement them when customers demanded them. Rippert realized just how important tools for handling the onslaught of unstructured data would become, and he decided he wanted to get into that space for a living.
He knew some of Basho’s executives, including then CEO Earl Galleher, from their time at Akamai, so they got to talking. It turns out Galleher wanted to transition to a chairman of the board position, so he stepped aside to let Rippert come in as CEO.
Now that he’s been there for a while, Rippert said there are some differences between Basho and his former employer. Sheer head count aside, Rippert is now immersed in the core technology, which is a big change from his role at Accenture. Further, he said, it requires a big attitude adjustment to move from a services company to a software company. Whereas services companies react when technologies come to them, he explained, software companies have to make a major investment in technologies long before the revenues start rolling in.
Basho has already made a lot of that investment, Rippert said. Riak is a fully distributed key-value store that’s based on Amazon’s Dynamo research paper, although all of its code was written from scratch. Now he thinks his industry experience will help him position Basho’s Riak as a legitimate option for enterprise data centers.
Forty percent penetration for NoSQL
Enterprises will start adopting NoSQL en masse, Rippert thinks, because the types of data they’re now dealing with require new technologies. “We are the data store for the new type of data being stored,” he explained. Existing relational databases were invented decades ago, he said, but they weren’t prepared for the worldwide data-volume surge that hit around 2004. And volumes are only growing — IDC predicts that more than 1.8 zettabytes, or more than 1,800 exabytes, of information will be created and stored in 2011.
That data is largely of the unstructured variety coming from web applications, machines and other sources that aren’t the traditional business-transaction data for which relational databases were created. Relational databases were the answer to almost everything previously, but now Rippert thinks NoSQL is “the answer to about 40 percent of business use cases today.” Relational databases aren’t going anywhere, but he thinks NoSQL databases can move in on new applications and replace existing databases that are storing data better suited to NoSQL technologies.
Of course, he acknowledges, NoSQL has nowhere near 40 percent of the database market. “When I go to sleep at night, I dream about 40 percent market penetration, but we don’t have it now,” Rippert joked. Although, he noted, NoSQL easily has more than 40 percent penetration among web developers and other folks building next-generation applications. And as those applications grow in number, so will NoSQL’s market share.
He thinks the tipping point for traditional enterprise adoption is about three years, after companies see more use cases pop up and become more confident with the products. Rippert noted that NoSQL databases are generally far less expensive than relational databases, so that will help, too.
While at Accenture, Rippert said one large technology client was already using Riak, as is Comcast as part of the infrastructure for its Xfinity cable service. Since joining Basho, he said the company has signed on one large customer and has several very promising leads.
Expect lots of M&A
Assuming enterprise adoption picks up like Rippert thinks it will, he predicts that legacy software vendors will start snatching up NoSQL startups to ensure they get a piece of the pie. He thinks integration likely will be a tougher task for vendors that also have a hardware business, though, because NoSQL databases are designed for scale-out, commodity hardware. Those companies will try to sell their high-margin gear as long as they can get away with it, he said. However, NoSQL does add some higher-level opportunities around analytics and visualization that will help offset those hardware losses.
If NoSQL companies are to survive on their own, Rippert thinks they will have to expand beyond their current capabilities. Customers will be willing to run multiple databases, he thinks, but the fewer they have to buy, the better. For example, he said, a NoSQL database that blends key-value, document and columnar features would be a good idea. Basho is working on expanding the capabilities of Riak, he added. Couchbase recently launched its flagship product that melds the Membase key-value store with the CouchDB document database.
At this point, Rippert said Basho spends most of its energy convincing customers that NoSQL is a legitimate option rather than comparing itself directly to competitors such as MongoDB or Couchbase. As I’ve noted before, though, as business picks up and large vendors start drooling over acquisitions, so will the competition among the still fairly congenial NoSQL crowd.