2 Comments

Summary:

Platfora, the San Mateo, Calif.-based startup that helped spur a general rethinking of business intelligence for a big data world, is finally exiting its beta period and is generally available. It’s no wonder the company has garnered so much attention given its stated mission to make […]

Doing Hadoop-powered BI with Platfora

Platfora, the San Mateo, Calif.-based startup that helped spur a general rethinking of business intelligence for a big data world, is finally exiting its beta period and is generally available. It’s no wonder the company has garnered so much attention given its stated mission to make Hadoop an interactive experience and to disrupt a multi-billion-dollar data warehouse and BI market.

Unlike legacy BI applications that generally connect to Hadoop but otherwise retain their old-school performance limitations, Platfora and its ilk have big data at their core. Platfora is built on Hadoop for scale, but the company also has its own IP around in-memory processing to improve the speed of slicing and dicing through data, and its HTML5 interface provides an easy way to navigate through lots of data points.

Justin Borgman Hadapt Tomer Shiran MapR Technologies Ashish Thusoo Qubole Ben Werther Platfora Structure Data 2013

Werther (far left) talking SQL on Hadoop at Structure: Data 2013 along with representatives from Qubole, MapR, Hadapt and Facebook.

I’ve compared this general family of products — in which I’d also include ClearStory, Precog, SiSense and Birst, among others I’m sure — to Tableau, albeit slightly (sometimes significantly) rethought and then jacked up on steroids to handle big data scale and/or speed. The big difference with Platfora, though, is that it’s built on top of Hadoop and is therefore part of an even bigger movement around that open source platform and a quest to build native SQL queries into a system designed for MapReduce.

We have been covering Platfora since its inception, from stealth mode to launch, and then a whopping $20 million VC investment in November.

  1. John A. De Goes Tuesday, March 26, 2013

    Hey Derrick,

    Platfora is among several companies in what I term the “big data BI” space. Most players in this space fall into the backend (Hadapt) or the frontend (Datameer), although a few companies span a bit of both worlds (Platfora). A highly successful end-game for Platfora is becoming the “Tableau of big data BI”.

    Precog is not focused on the BI market, because its needs are well addressed by existing players in the space (and like most spaces, there’s not enough room for more than a couple major winners in big data BI).

    Precog is firmly focused on data science, i.e. the tying together and deep analysis of heterogeneous data sets with varying levels of structure, quality, data policies, and size, for purposes of finding revenue-generating data insights and building revenue-generating data products.

    Companies want to perform deeper analysis of ever-more diverse data sets consisting almost entirely of non-relational data, and they want to use these analyses to power their business. Netflix, for example, uses data science to deeply understand user behavior and build recommendation and personalization features into their product.

    BI is not designed to solve these problems.

    While big data BI is still critical to businesses and will no doubt be successful, there is an emerging, rapidly growing market for companies who want to use more types of data to perform far more than just BI. And that is the market that Precog is singularly focused on.

    1. Hi John, Precog looks like a nice solution, will sign up for the developer programme and get a more detailed opinion :)

      In my view though, we are only starting on the analytical path of big data. In my view all data will form part of this, structured and unstructured, and will be seen as one big data entity. The tools to mine through this data will also evolve from a reporting function that it is today to a semi-automated decision function in future. For instance, all business processes will be managed from a big data perspective, and when there is a higher than usual exception activity on a queue, the system will give a number of possible solutions, which might even entail the change of a business rule to better deal with the exceptions.

Comments have been disabled for this post