I noticed in some posts I wrote in the past the different ways to calculate the number of connected objects. Telecom based companies (like Ericsson, Cisco) tend to estimate in around 50 billions the number of connected objects by 2020, Information Technology (IT) based companies (like HP, SUN) tend to estimate in 1 trillion the number of connected objects by 2020, that is a 1:20 difference!
Now, I stumbled upon the estimate provided by IBM in its 2013 report that is also estimating 1 trillion connected devices generating data, but the estimate is for 2015!
This estimate from IBM is coupled with the fact (according to them) that every day the "world" is producing around 2.5 Exabyte of data, most of this in an unstructured way, that is in forms that are difficult to process. Actually, IBM estimates that in 2014 only 1% of the overall date produced are used to derive information. The rest is simply disregarded. And yet, there is plenty of information hidden in the remaining 99%.
The problem is that the processing of the amazingly big amount of data requires much more processing that what is available in today's big computers.
The answer to this increasing need for processing should be met by different computing architectures. For IBM the answer lies in data centric processing. Rather than moving data to be processed in big computers it is more effective to move processing capacity to where data are stored.
This change of processing paradigm is also pushed by the US Department of Energy that has asked IBM to develop the most advanced supercomputer with a data centric architecture, with a processing capacity of 150PetaFlops, that is 5 times bigger than today's fastest supercomputer (Thiane 2).
The intention is to be able to process a bigger slice of the data pie. This according to IBM and DoE should provide significant improvement in areas as diverse as hurricane path prediction and health care. For this latter high computation capabilities would allow the emergence of cognitive computing, an area where IBM is investing over 1 billion $ next year to bring Watson capabilities to medical doctors (Watson is the computer that won Jeopardy in the US).
Indeed, being able to process big data, in genomic, drug design, statistic monitoring of epidemics, may well be a substitute of the famous "an apple a day keeps the doctor away".