Platfora, the Hadoop-based startup from Greenplum and DataStax veteran Ben Werther, has raised $5.7 million from Andreessen Horowitz and intelligence-focused strategic investor In-Q-Tel. Investors are excited because Platfora promises big things around making big data analytics obtainable by anyone needing to parse large volumes of unstructured data, not just data scientists.
Hadoop has proved itself very effective for storing and processing big data workloads, but writing applications atop Hadoop’s MapReduce framework, or even doing queries with its SQL-like Hive overlay, is no easy feat. Companies such as Cloudera and MapR have raised a lot of money to improve the management and performance of Hadoop clusters, but they don’t address higher-level concerns such as creating applications. As Platfora founder and CEO Werther puts it in a recent interview, Hadoop is just half a solution: “It’s the plumbing.”
Werther said Platfora wants to make Hadoop functional to the point that even a history major could perform meaningful analysis across mountains of data. It aims to do that with an intuitive user interface that has advanced data science functions built in, rather than making users perform queries.
Other startups, most notably Datameer, also aim to make Hadoop analysis easier for business users rather than data scientists, but Werther says there’s a big difference between what they do and what Platfora will do when its product is available next year. He said Platfora’s interface will be easier to use than even Datameer’s spreadsheet-like approach, and it will be more interactive because it will give users sub-second response times for their results. Hadoop is typically a batch-processing framework, meaning users have to wait several minutes or longer for results.
Aside from making Hadoop more palatable for mainstream organizations, Werther also thinks Platfora stands to shake up the data warehouse and business intelligence markets. Presently, data warehouse architects determine the data worthy of being stored, meaning that business users are limited in the information they have to query against. Doing business intelligence on top of that data requires an additional software product, and working with unstructured data stored in Hadoop requires a connector between the two environments.
Werther thinks Platfora can mitigate the need for all of these products because it performs all of these functions to some degree. And because Hadoop is so scalable, companies don’t have to limit the data they choose to retain but can keep it all. The Platfora software, he says, will let users easily determine the parameters of the data they want to analyze.
It all sounds great on paper, but there is at least one catch to Platfora: Customers will still have to employ at least a few Hadoop-savvy folks to manage their Hadoop clusters. Werther said that most early adopters already have Hadoop deployments in place, so Platfora will initially be a software product that runs atop those clusters. At some point, though, he noted, Platfora might integrate Hadoop into its product to make it a more holistic experience for mainstream users without existing Hadoop expertise.