Once a big-data product starts gaining adoption, the next step for promoters is to help it go mainstream by adding all the deep management and security features that enterprise IT shops want. 10Gen reached an important milestone on that front on Tuesday, announcing that it has secured $42 million in new venture capital financing. It's an infusion of cash that will fund a lot of development work.
The latest round of financing, 10Gen's fifth since the company's launch in 2008, was led by New Enterprise Associates with participation from existing investors Sequoia Capital, Flybridge Capital, and Union Square Ventures.
It's a vote of confidence in a product that 10Gen claims holds nearly half the NoSQL market. Definitive market share statistics are hard to come by, but the company points to research by the 451 Group that shows that MongoDB has three to four times the number of developers, some 12,000, than competitors Redis, Apache (Hadoop) HBase, CouchDB, or Apache Cassandra.
[ Want more on NoSQL databases? Read HBase: Hadoop's Next Big Data Chapter. ]
10Gen says it has more than 500 paying customers on its subscriber-edition of MongoDB and more than 5,000 organizations are using its free MongoDB Monitoring Service (MMS) to scrutinize subscriber and open-source deployments of the database. 10Gen differs from support, training, and consulting organizations, such as Cloudera or DataStax, in that it is also the developer of the open-source product that it supports.
"Some big corporations are still nervous about open-source software, but because we created the database and own the code, we can make it available both through both open-source and commercial licensing," said Max Schireson, President of 10Gen, in an interview with InformationWeek.
GNU-licensed MongoDB has a strong open-source community, says Schireson, but there are signs that community could ultimately be eclipsed by that of HBase, the NoSQL database within the Hadoop Framework. With Cloudera, Hortonworks, MapR, and others distributing code and otherwise supporting Hadoop, HBase can feed off a large and competitive community while also taking advantage of the ever-growing data stores held in Hadoop deployments.
Like other NoSQL databases, MongoDB gives users the flexibility to store and recall any type of data without the rigid constraints of columns and rows. That means you can add new data types--including complex data and loosely structured textual information--without conforming it to a predefined schema. In contrast, the schemas behind conventional relational databases such as IBM DB2, Microsoft SQL Server, MySQL, and Oracle database have to be revised with each change in dimensions.
NoSQL databases are a favorite with businesses such as large-scale Web companies that deal with complex and variable data sources such as clickstreams, Web logs, and social-media data feeds. 10Gen's customer base includes the likes of Craigslist, eBay, Foursquare, Shutterfly and Wordnik. But as MongoDB has matured, it has also gained traction with mainstream media companies such as Viacom and Disney, telcos such as Telephonica, tech giants Cisco and Intuit, and even government agencies in the U.K. and India.
At last week's packed MongoNYC event at the Pennsylvania Hotel near Madison Square Garden, I spotted a few representatives of big banks (like Goldman Sachs) among the more than 900 people in attendance. The list of customers presenting included speakers from The New York Times and Forbes, who touted MongoDB's versatility and ease of use during a panel discussion about online media's use of the database.
Kapadia demonstrated Cascade, a MongoDB-based application now in production use at The New York Times. The app tracks every Twitter tweet and retweet tied to each story published online by the news organization. Analysts, editors, and writers can then use a visual analysis interface to track traffic trends and uncover influencers.
For example, Cascade uncovered that tweets tied to the story "Mexican Youth March Against Old Ruling Party" skyrocketed after a well-known Mexican columnist with more than 100,000 followers tweeted about the story. This helps the news organization follow the hottest topics and spot potential sources tied to key issues.
To provide this insight, Cascade tracks the intersection of the fire hose feed of Twitter tweets and retweets with a custom Bit.ly stream of URLs for NYTimes stories. This big-data stream of all tweets and retweets related to times stories is then presented in tree-and-graph data visualization that shows trending topics and the tweets that spawn the most retweets.
[ Want more on MongoDB in action? Read 2 Lessons Learned Managing Big Data In Cloud. ]
MongoDB is "far easier than managing a MySQL cluster or an Oracle cluster," Kapadia said, noting that NYTimes Labs doesn't need a dedicated DBA to manage the product--developers can handle required administrative tasks in their spare time. In massive production settings, MongoDB administration might be a part-time job, Kapadia acknowledged.
Next steps for MongoDB include a 2.2 release set for this summer that will improve control over database concurrency. The release will also introduce an aggregation framework, for better ad-hoc query support, and location-aware sharding, a feature needed in multi-data-center deployments. All of the above are capabilities that 10Gen's most demanding customers are asking for, Schireson said. And it's the kind of development that this week's infusion of cash will support.
With Oracle and Amazon wading into the NoSQL game with the Oracle NoSQL Database and DynamoDB service, respectively, there are clearly bigger fish in the market--which is growing 21% per year and is expected to reach $3.8 billion in annual software and services revenue, according to Market Research Media. But Schireson says that the Oracle and Amazon products are closer to simple key-value store-style NoSQL databases that don't compete head on with MongoDB.
"What we support that key-value stores don't support is querying against lots of different fields," Schireson said, citing the example of an order management system. "With a key value store, you could query for a particular order, whereas with MongoDB, you could query for all orders placed by a particular customer that are more than a week old but that haven't shipped yet."
10Gen is a dwarf compared with the likes of Oracle, but with NoSQL adoption on the rise and MongoDB riding the wave, Schireson says he's optimistic that many more enterprises will give the database a try.
"Oracle had enormous resources and for the last couple of decades they've been a kind of default choice," said Schiereson, himself a former Oracle employee. "We'd like developers to think about what's the right fit for their project before they select a technology, and if they do, we think a lot of them will select MongoDB."
The database that has the most to lose from MongoDB and other NoSQL databases is MySQL, the open source database now owned by Oracle. MySQL has a large and dedicated following of developers, but it's a scale-constrained conventional database that's now growing much more slowly than the burgeoning, big-data friendly NoSQL community.