Wonderful. On same lines, here are few more posts from the past. But our industry is notorious for snake-oil peddlers who sell it to the gullible in many shapes and forms.
2009 - Joel Spolsky - The Duct Tape Programmer
2010 - Ted Dziuba - Taco Bell Programming
2013 - Chris Stucchio - Don't use Hadoop - your data isn't that big
2013 - Ashish - Don't Use Hadoop. Your Stovepipe Still Works
2014 - Adam Drake - Command-line Tools can be 235x Faster than your Hadoop Cluster
2016 - Randy Zwitch - A Million Text Files And A Single Laptop
2017 - Alex Woodie - Hadoop Has Failed Us, Tech Experts Say (On Slashdot)
This is not how rational people make decisions, but it is how software engineers decide to use MapReduce.
But it is not that simple, IMO. Deep down, most of the decision makers crave to sound profound and important. And in that quest, they end up complicating things. In big organizations, getting budgets to rig up and run SOA/Agile/BigData/Analytics/DevOps/AI/ML/VR/AR teams is easily seen as being more profound and accomplishing than getting things done the old-school ways. For startups, it's the classic shiny object syndrome since they would be forced to lure investors (and potential hires and employees by selling the coolness quotient). In order to appear cool, few end up drinking the kool-aid, at times excessively.
Real engineers would see through the farce and wherever possible (and to whatsoever extent possible) would try to fix things. If they can't, they figure out a way to maintain their sanity and move elsewhere. And the clueless will carry on with the same cycle, over and over.
