Chowdhury takes an interesting stance on big data, opining that analytics needs to evolve into something he calls ‘fast data’ and ‘smart data’. He emphasizes on the scaling of big data infrastructure, and how the natural evolution of big data is ‘smart data’ at CIO Year Ahead 2017.

“We’re going to be talking about the convergence of data. It’s a slightly different concept from the big data that we’ve so far known. Big data is all about curiosity – to find the connections between seemingly unconnected factors,” says Chowdhury.

Do analytics at the edge – that’s where you find the beauty in the data. Chowdhury coins a witty term for the manner in which big data is being used nowadays – he calls it “Hanuman engineering”. It’s when you don’t know what you exactly need, so you bring in all the data you could want.

Smart data, on the other hand, is big data turned into actionable data. It’s available in real time for a variety of business outcomes, whether it’s in industrial applications, data-driven marketing, or process optimization.

Citing examples of applications where big data is not used smartly enough, he points at vehicle and asset tracking. “Once the vehicle has been dispatched, you do not need to keep track of all the information it generates,” he says.

Another example is utility metering – 99.9 percent of the time you don’t all the data that’s generated every 15 minutes. You can install a single machine costing less than 1000 rupees, which generates precisely the data you need. Now that is smart data.

Addressing another concern plaguing CIOs, Chowdhury points out at where one embeds the intelligence. “Do you embed in the device, the gateway, cloud app, or the app itself?” asks Chowdhury.His take is that the answer lies in the manner in which you design – good designs allow you to capture the data only at the edges.

“You also need to decide if you want to stream the data or store the data. You also need to figure out if you need to throw away that data, and just store the intelligence you derive,” he says.You need to look at real time computing, distributes computing, and fault-tolerant computing.

Speaking about fractal mathematics – an area Chowdhury believes is elementary when it comes to addressing the characteristics of data, he says, “The characteristics of these help you predict the nature of outcomes. The more iterations you do, the more replications you find,” he says.

Chowdhury shares his precious insight when it comes to recommending where the analytics needs to be done. “Do analytics at the edge – that’s where you find the beauty in the data. That’s where you’ll find the really interesting data,” he says.