Look Out, Mongo, Here Comes Couchbase
In the battle for NoSQL supremacy, there will be winners and there will be losers. MongoDB got off to an early start and grew quickly through a combination of big rounds of venture funding and customer license deals. But according to Couchbase CEO Bob Weiderhold, his NoSQL database business is just starting to ramp up, and now he’s gunning to take market share from Mongo.
The NoSQL database market grew considerably in 2013, as big established companies decided that the technology was mature enough to be used in critical business systems. According to Wikibon’s Big Data Vendor Revenue and Market Forecast 2013-2017, the NoSQL database business accounted for about $290 million in 2013. That’s just a small part of the overall $35 billion database business. But it’s one of the fastest growing segments, with Wikibon expecting 74 percent market growth in the NoSQL segment this year, making it a $500 million business. (That’s considerably bigger than the Hadoop business, by the way.)
Couchbase is positioned to capture much of that NoSQL growth, according to Weiderhold. “Coming into the year we were already recognized as leader in NoSQL with Mongo and Datastax,” the Couchbase CEO told Datanami. “But we grew much faster than they did in 2013. As we see this trend continuing in 2014 and accelerating, we think we’re going to grow faster than them this year as well.”
Much of the 400 percent growth that Weiderhold claims Couchbase had in 2013 can be attributed to replacing relational databases from Oracle, Microsoft, and IBM. When these big Web and mobile business systems were originally developed in the late 1990s and early 2000s, relational technology was the only game in town, so that’s what was used.
“Now they’ve come to conclusion that NoSQL is mature enough, and that in order to support the huge number of users, in order to support huge amounts of data, and in order to support the usage patterns, they really need to break from relational and move to NoSQL,” Widerhold said. “That’s a key part of the trend. They’re coming to the conclusion that it’s mature enough and they’re making a strategic decision to begin moving to NoSQL, and then they’re doing these broad deployments.”
Couchbase won the majority of the product NoSQL product “bakeoffs” it was involved in, Weiderhold says. “Our real strength from a differentiation perspective is scalability and performance, compared to Mongo in particular, and Datastax as well,” he said.
|Couchbase CEO Bob Weiderhold|
Weiderhold pulled no punches in discussing his competitor MongoDB. “It’s very difficult to scale with Mongo and oftentimes there are reliability problems associated with scaling,” he said. “Quite frankly their architecture for scaling is dramatically different than our architecture for scaling. We have a single node type with no single point of failure. You can literally press a button and add additional servers to your cluster and the cluster will automatically shard your database. It’s very dramatically easier to scale using Couchbase than it is using Mongo.”
MongoDB, of course, would vigorously deny that it has trouble scaling. “MongoDB is built for scalability, performance and high availability, scaling from single server deployments to large, complex multi-site architectures,” the company states on its website.
However, last fall, the company was taken to task for a post on its blog about the limits of scalability. “Surpassing 100GB of data in your application requires you to have in-depth knowledge of how to operate and run MongoDB,” read the opening sentence in the blog post, which was taken down but cached by Jaxenter.com. “MongoHQ [a sister company to MongDB] recommends going through the 100GB Scaling Checklist as you grow.” In most circumstances, 100GB would not be considered “big data.”
Weiderhold did not take any prisoners in his blunt assessment of the technical limitations of his older, bigger, and richer competitor. “Mongo has four to five node types. They have single points of failure. They encourage customer to do range-based partitioning of data across a cluster,” he said. “Those kinds of things cause significant additional complexity in terms of scaling, and cause significant additional problems in terms of reliability. Those are the things that people are finding when they do deep strategic evaluations. On the performance side, they key thing for us is we have a managed cache. It is Memcached. We have very fine-grain locking…so it’s a very small amount of data that’s locked. Mongo has course-grained locking and that causes big performance problems for them.”
Couchbase is clearly positioning itself as the alternative to MongoDB. Both have similar architectures, which are combinations of key value stores and document databases. Both Couchbase and MongoDB use JSON as the data model. Datastax–the other leg in Weiderhold’s three-legged NoSQL stool–is different in that it’s a column-oriented data store.
The folks at Datastax do not subscribe to the theory that Couchbase holds title to the world’s best scaling NoSQL database. “Keep in mind that Couchbase does not scale like we do and they admit that,” Robin Schumacher, DataStax vice president of products, told Datanami. “Matt Aslett at the 451 Group just came out with an update on Couchbase and they admit in that report that they don’t scale like we do. They said that they’re for different types of use case. They’re positioning themselves as a mobile type of database and documented oriented database, where we are more of a larger scale database that is geared toward online applications, time series data, and lots of it.”
Weiderhold doesn’t deny that. “The way we think about it, we’re focusing exclusively on mobile and Internet applications,” he said. “If you were developing a mobile application between 1995 and 2002, you had no choice but to develop it using an Oracle database or IBM DB2 or Microsoft SQL Server. We’re seeing many companies that developed on those database feel like NoSQL is ready for prime time. It’s mature enough, and they see huge advantages, so they’re switching off those technologies onto NoSQL, and more specifically Couchbase.”
While MongoDB has run away with the venture-funding title (Bloomberg estimates its worth at more than $1 billion), the Couchbase CEO says its 400 percent growth rate in 2013 shows that it’s winning. And while Oracle, IBM, and Microsoft provide relatively soft targets for all NoSQL vendors to work against, Weiderhold is now looking to poach customers from Mongo, too. Its recent customer win at online phone company Viber was the result of a 8x performance boost it got from switching from MongoDB to Couchbase, the company says.
It’s all about winning, Weiderhold says. “Those are big deals for us. That’s what we’re winning,” he said. “We’re winning those strategic evaluations. We’re doing these big deployments and getting big deals as a result of winning these and that’s why we’re growing so much faster than MongoDB or Datastax, because we’re winning them.”