Some Comments on the Five Fallacies of Cloud Computing

When searching for the five fallacies of network programming, I came across the Five Fallacies of Cloud Computing. A short digression on search. Search engines are contributing to the “Distraction factor”, you look for x and are shown y and you spend time on it. I wonder how much time Google and Bing actually cost us with irrelevant hints that we have to filter. The fact is they are still better than going to the library or looking it up in a book. Back to the five fallacies of cloud computing. I found that only one of them had strong scientific merit and that was number 5, the cloud is secure. Cloud security is only as good as your service provider —  there are no solutions at the current time that allow you to perform remote secure distributed computation, see previous post on Homomorphic Computing.  If I put an ICE unit on any processing unit I can extract clear text with current technologies. Homomorphic Computing would survive that type probe. Recent advances in Quantum computing may provide an alternate path. for secure remote computing.  Quantum computing would also survive the ICE probe due to the Heisenberg uncertainty principle. The article makes some fine points about that basically are about.

  • New technologies almost never totally replace old technologies, what is important is the percentages not that some data centers will remain.
  • IT has many applications that are too complicated to unwind, that doesn’t mean that it can’t be done and is not cheaper in some cases.  Kind of like the issue with energy inefficient homes, costs to much to fix them with the current costs. If I were a CIO or CFO and I was told that my app is too complicated to move, I would be very concerned about how that app is implemented. Perhaps the apps need to be looked at more carefully.
  • I am always amused with the difference between coming up with an idea and actually making it work, two totally different things. Engineering is about ideas that can be realized and made reliable in some limited time frame.
  • Often new technologies don’t succeed immediately but experience an S type curve of adoption, Kurzweill has an interesting argument, Kurzweil’s Law (aka “the law of accelerating returns”) attempting to technology adoption is actually a bunch of stacked S curves moving forward in time on the x-axis  and adoption and capability on the y-axis
  • Nothing is really new, I find this to be a lack of appreciation for incremental progress and features. Products often improve and in many cases are better than the previous version. This is one of the reason why I like  marketing people and college professors, they often appreciate the subtle differences that are actually improvements. This is similar to the generational thing when some don’t get it quickly and adopt it later, but fail to provide those that do get it the benefit of the doubt.

IMO the fundamental challenge for  Cloud Computing is security and bandwidth to and from Public Clouds, this is not as important  in Private Clouds.  Perhaps Homomorphic Computing and Quantum computing will provide implementation approaches. I don’t know enough about Quantum computing to understand how it could help with the Bandwidth issue. In the short term, I will keep working with all the current useful tools and approaches to efficiently managing a bunch of resources.

A similar movement NO SQL has an analogous problem with failing to appreciate and truly understand classical Computer Science  facts like ACID.  Another movement had a similar issue, The Semantic Web. The Semantic Web movement is just recently realizing that NLP and Reasoning Systems are extremely difficult challenging Computer Science problems. I applaud and participate in many of the above efforts and think it would be better if  there was more honesty in explaining the limitations of each of the technologies at the current time.

Related Posts

Leave a Reply