Big Data Analytics can be so confusing. This is not an infant industry, for heaven’s sake. 

Let me give you a view of my working week.

Monday: went to an online betting organization. Death would occur in the business if they could not identify the source and value of new customers.  Death, you understand. Not mere discomfort or searing pain, but death! The only antidote, apparently, is the production of visual representations of their customer flows, with leakage points and value drivers popping out of the eye candy to alert the business where to change its marketing or operational behavior.

Tuesday: Travelled far across ‘a great and impassable desert’ to an online shopping company who want to know how many customers are spending and on what, in a series of weekly reports. But, to delve deeper, they now need a modelling engine to predict and intercept activity on the website.

Wednesday: Jousting with two media buying organizations to create online trigger platforms for their seriously ‘whizzy’ model builders and bid management engines. They were challenging our ability to calculate predictions and execute them according to the latest onrush of vast and unstructured log files and tags. I showed ‘em.

Thursday: Conference calls with US Medical organizations: MAKE MY REPORTS GO FASTER!!! I WANT THEM NOW!!! NOW, YOU HEAR…..Oh……thank you. (Smug smile.)

Friday: A WebEx with an analytical services company creating alchemy in the digital space with sloppy truckloads of unrelated data and turning it in to real-time decision gold.

My point?  It’s so varied, out there.  All of it called ‘analytics’, with the data ever growing and changing, and the degree of ‘analytics’ determining the price the customer pays for the analytical platform (and the services come to that). That bit, the price, seems harder for them to grasp. How hard can it be to report data? So, our job in the industry is to simplify this for them and make the cost acceptable. We have to make it easy to collect data, make it easier to dump it in to a single place and make it straightforward to extract the bits you want for ‘analytics’. This is what I deal with day out and day in here at Kognitio; fitting diverse requirements to diverse budgets on a single platform. Luckily for me help is at hand.

Let’s take an analogy. Plastic building blocks from Denmark (I’m scared of brand names). You have a big bucket of blocks. They are in no order, you scoop a handful out and select the pieces you want and throw the rest back. Or, you peer inside hunting for the pieces you want. You build your ‘model’ (I love this analogy) and you still have a jumble of data, sorry, blocks to do more. Wouldn’t it be great though if you had something that just scooped out the right bits, really, really quickly and placed them neatly in front of you?  Well, guess what?….It exists! For data, I mean (nearly got carried away there).

End of analogy, return to the real world of Hadoop and In-Memory, Massively Parallel Processing. Big data (blocks) in a big bucket, largely unstructured, is what these customers are faced with. Rapidly expanding disparate data with as many analytical questions as there are people to think of them. The clever part is emerging in the tools and services to deal with this. Tools, such as those provided by MetaScale, will permit you to throw your data into the big bucket and haul out the bits you want, when you need them. Conveniently they hand them to an analytical platform with in-memory, massively parallel processing capability to make the ‘analytics’ fly (Kognitio). The data flies in, it zooms out and the analysts dart and dive through the data, free of traditional database bottlenecks.

So what does that mean for our customers and their analysts? Well, to begin with it means I can talk to them about something interesting, so they don’t get bored when I go and see them. But most of all they can control the budget (services like this are cloud-based), expand their data use without impacting current systems, and do ‘analytics’ (whatever that means to them) from weekly reporting to Vulcan thought programming without resort to a confusing array of systems.

Lucky them!

Nigel Sanctuary
VP Cloud Propositions, Kognitio