I am writing this the day before traveling “into the teeth of the storm” as they say. This should be interesting because the dreaded “Frankenstorm” is bearing down on the U.S. East Coast just as I need to travel there to teach a sequence of courses for Learning Tree, their Linux Server Administration and Support course followed by their Cloud Security Essentials course.
I was scheduled to fly but there is concern that airports may be shut down, or at least many flights may be canceled or significantly delayed.
This isn’t the end of the world. They called me four days in advance and asked if I would be willing to change my plans and fly to Washington a day or two early? Or maybe drive?
I decided that if I had to change plans and lose a day or two, let’s at least make it interesting. So I’m taking Amtrak from Indiana to the East Coast and will return the same way. Yes, this involves an overnight train in each direction.
What is the cloud (computing!) angle here?
Imperfect availability can be acceptable!
It will be interesting to watch how this storm turns out. I feel pretty safe in predicting that this storm will not lead to widespread outcry that the idea of aviation is terribly flawed. No one is going to seriously advocate that everyone give up on commercial airlines, let’s go back to traveling exclusively on trains and by road. There will be little sympathy for anyone who ignores a hurricane until an hour before landfall and only then realizes that they can’t fly into the affected region.
So why is the cloud held to an unrealistic standard?
The main point: The available infrastructure is redundant and distributed. Use that to your advantage!
Meanwhile, so many of Amazon’s customers are fixated on the US-East-1 region, which was battered by the derecho last June and has much of its infrastructure in the landing zone for some hurricanes. It’s not just that many use that geographic region primarily, but many use it exclusively.
Read those Amazon recommendations, people!