Big data, advanced analytics, machine learning, stream analytics… these are buzzwords that organizations believe will completely transform the way they do business. And they are right! The business value is somewhere locked behind the large volumes of structured and unstructured data coming from inside and outside the company. To unlock it, these organizations must be able to store this data, process it, identify trends, detect patterns, and find actionable insight in the sea of available information [CLICK TO TWEET].
In this context, the choice of the right technology is crucial. Indeed, most organizations today are spending millions of dollars in installing and experimenting with very expensive hardware (often badly sized), and spending thousands of hours building skills and implementing solutions that often don’t even work like they should, thus not exploiting the entire potential of the product.
Is there another way to do this? Absolutely! Follow these two principles: make sure you do the right things, and make sure you do them the right way.
Make sure you do the right things
We often see organizations having a big debate on which technology to use before even knowing what their analytical needs are, or they adapt their needs depending on what the technology they are planning to use offers. These are the first signs that their big data strategy is very likely to be unsuccessful.
The first thing to do is to have a clear understanding of the needs and capabilities you want to have when it comes to starting your journey towards big data. And more importantly, this should be regardless of any technology you’re planning to use.
Do how do you get started, especially when you have no idea on how to approach these concepts? The answer is obvious: a Proof of Concept for a specific business case, using a low-cost but powerful technology that you can activate whenever needed and deactivate whenever no longer needed. This may sound like a dream but many vendors, like Microsoft with their Cortana Intelligence Suite, are already proposing these solutions in the cloud, and are adopting a pay-per-use strategy.
Make sure you do them the right way
Now that you’ve experimented with these concepts and capabilities, you are ready to clearly define your analytical needs and the capabilities you will need to deploy.
Once you define your needs, evaluate the different technologies and/or vendors with regards to the solutions they propose to answer those needs with. Don’t forget that as long as we are talking about big data, advanced analytics, data exploration, machine learning, and so on… the preferred approach to do this type of work is agile, with “try and test” iterations. Therefore, the technical solution and architecture should be flexible enough to support this way of working. And here again, the cloud and pay-per-use solutions should be your best friends.
This sounds like a good theory but how to plan and execute such a Proof of Concept? I’ll discuss this in my next blog so stay tuned!