Meet the New Cloud

Roland is a member of The Motley Fool Blog Network -- entries represent the personal opinion of the blogger and are not formally edited.

Same as the old cloud

Sorry, I couldn’t resist ripping off The Who “Won’t Get Fooled Again," nor could I resist screaming that title out as I typed it. Thankfully, I work alone, although after reading the opening to this it might be a bit difficult for you to believe that is by choice.

If you haven’t already read my previous post on the old clouds, please do so:

http://blogs.fool.com/SeasonedGeek/2011/12/13/anybody-old-enough-remember-last-cloud/

Every cloud vendor on the market will be quick to point out that their technology has nothing to do with frame relay. While that would be true, it has nothing to do with why they are the same. The old cloud looked good on paper until greed got in the way. The new cloud looks good in an MBA spreadsheet, until greed and reality get in the way.

Google (NASDAQ: GOOG), Microsoft (NASDAQ: MSFT), Amazon (NASDAQ: AMZN) and other vendors got into this game promising to offload all of the central computing hardware needs of corporations into “the cloud.” All they had to do was convert all of their applications to run on either Linux or Windows and “the cloud” would be theirs for the taking. It was a true pay-as-you-go model. You will only be charged for the CPU cycles and data storage you actually use. You no longer have to worry about updating computer hardware or software for your data center. That BRP (Business Recovery Plan) you have to write and test to be SOX compliant, it’s now a one-page document that says “buy new PC, connect to cloud.” We’ll handle everything else.

Sound familiar? Anyone care to look up how 1970s time sharing worked? Deja vu all over again. Remember how the last cloud was going to magically increase bandwidth on demand? Well, this cloud is going to magically increase your CPU count on demand...for a small fee.

The fatal flaw with this new cloud is once again greed and reality. Most, if not all of these companies built a massive data center someplace where they had both cheap space and cheap power. ALL of their redundancy is inside of this data center, which, in most cases, is a single building without hardened security. From the pictures I’ve seen, many of these data centers appear to be in business campuses where anyone with a rented van full of diesel fuel and fertilizer can park right next to the building just like happened in Oklahoma. Of course, as most of us learned earlier this year, to bring eCommerce to a halt for cloud users, you don’t have to go that far.

http://siliconangle.com/blog/2011/09/09/microsoft-google-cloud-outages-affect-millions/

You see, my background is in designing great systems built around real operating systems. The ones that actually cluster, not the ones which make false claims of the capability. (If the kernel of your operating system doesn’t have the definition of a logical record built into it, you don’t have the fundamental building block for a cluster, but that’s a topic for another time.)

Years ago there was a company named Digital Equipment Corporation (DEC). It was later consumed by Compaq which was then consumed by Hewlett Packard (NYSE: HPQ). They created an operating system that was first named VMS, then named OpenVMS for reasons that still defy logic today. It was the birthplace of the cluster and system designs that scattered tiny data centers around the globe. Physical hardening wasn’t needed because as long as a single location survived, you were still in business. This design, which the greedy and the cheap called overkill, proved itself during the September 11 attacks.

http://www.computerworld.com/s/article/97032/OpenVMS_survives_and_thrives

If you or anyone at your company thinks having all of your replication inside of one building or encapsulated by one multi-state power grid, then, perhaps you need to read up on the Northeast Blackout of 2003 and calculate just how long it will take to coordinate diesel fuel shipments from four states away. What? You say your data center backup generators are hooked up to “city gas”? Then you really need to read up and think about what happens when the gas company doesn’t have the power to run its pumps.

http://www.ask.com/wiki/Northeast_blackout_of_2003

I actually wrote about this very topic several years ago in “The Minimum You Need to Know About Service Oriented Architecture” ISBN-13 9780977086665.

This is a national security disaster of biblical proportions waiting to happen and we need to shut it down now. What happens if 60-90% of all banks subscribe to the same cloud? Believe it or not, an even worse and more likely scenario would be 70-80% of fleet dispatching/long haul trucking companies. One earthquake, one Tsunami, one insert-disaster-here that takes out the one data center housing that cloud and we’ll have fist fights nationwide over the last quart of milk in a store.

Don’t believe any cloud could achieve that level of market saturation? How about only one customer? What if it was AIG, Citigroup, or any of the other companies on Mr. Geithner’s “Too Big to Fail” list? What is the difference between failure and a three plus week long outage?


Roland Hughes is the President of Logikal Solutions and author of many titles.  He does not knowningly own a position in any of the companies mentioned.

http://www.theminimumyouneedtoknow.com
http://www.infiniteexposure.net

blog comments powered by Disqus