UT Dallas > Computer Science > Blog > CLOUD COMPUTING, FOG COMPUTING, AND BIG DATA: LESSONS FROM THE FASHION INDUSTRY

CLOUD COMPUTING, FOG COMPUTING, AND BIG DATA: LESSONS FROM THE FASHION INDUSTRY

By Dr. Ravi Prakash, UT Dallas

Every thirty years or so corduroys become fashionable. Raid your parents’ old suitcases and grandparents’ trunks. You will find that the clothes that were fashionable when they were young are in, once again. One can either attribute this to a lack of imagination on the part of the fashion industry when it comes to designing new clothes, or a genius on their part in terms of marketing the same old ideas generation after generation, or (and this is the more likely explanation) both. The field of computing seems to have paid close attention to this subterfuge.

The earliest modern computers had all necessary components in one place: the processor, primary and secondary memory and I/O devices. As technology advanced, a key I/O device, the terminal, was moved into another room, with a rather simple wired connection to the computer. When personal computers and workstations came along, we marveled at how powerful they were and wanted to perform all computation and data storage locally. But, soon, our computation and storage needs exceeded the capabilities of these desktop devices. So, we moved computation and storage to machines in other parts of the building, using our desktop devices as clients: thus was born client-server computing. Not content with just one marketable term, we came up with other impressive terms to inhabit this ecosystem: network file systems, thin clients, thick clients, storage area networks, etc.

With smartphones and tablets at our disposal, we have recycled the same game plan. Demonstrating that we have learned from the fashion industry (you may have the same basic idea, but you can’t call it by the same name), now we don’t have client-server systems; we have cloud computing. But, clouds are too far up in the sky, so to say, and it takes a long time to get to the cloud and back. So, we will tether these pesky clouds and pull them closer to the ground, and call them fog. I guess caching, which was cool until recently, has fallen out of favor. I presume soon we will also have mist, waterfall (sorry, software engineering folks have already appropriated that), nebulizer, atomizer, etc.!

This brings me to another revolutionary computing idea that will change the world (or not): Big Data. Big Data will change how we do everything from making simple purchases to choosing our life partners, to raising our kids, to believing in God. You may not believe me, but big data is indeed so big that it is even mentioned in the Bible. You do know that Jesus was born in a manger. Surely, Joseph and Mary were not homeless. So, why the manger? You may recall that the Gospel of Luke says something about traveling for a census (which is a very unfashionable term for a major Big Data exercise, coined by folks who were definitely not as cool as we computer scientists). Every nation conducts census periodically. Every ten years or so, countries like China and India gather lots of information about over a billion people each. This information is then used for allocation of tax revenue for various infrastructure projects, education and health initiatives, and, in the case of India, to draw electoral maps for the largest exercises in democracy. But, census is too old-fashioned a term and can’t be as important and consequential as something as cool and new-fangled like big data, right? Big data is indeed so big that we can stop teaching our students old-fashioned probability and statistics; they can forget about working with small, randomly sampled, representative data sets like that silly Nate Silver does. Just brute force solutions running in the cloud (yeah, that cloud, in case you have the attention span of people immersed in social media and can’t recall stuff in the previous paragraph) will suffice. Also, while we are doing all our big data stuff, we don’t need to study arcane concepts like file systems, memory hierarchy, I/O systems even though big data applications spend a good portion of their time actually moving data back-and-forth between main memory and secondary storage.

So, yeah, let’s all be very fashionable. As long as we look cool to the rest of the world, everything will be alright!