Thursday, September 29, 2022

Resilency II

 

Be Prepared

I know it sounds sordid
But you'll be rewarded
When at last I am given my dues
And injustice deliciously squared
Be prepared!

It is smart to be prepared!

A hundred year flood does NOT mean that a flood will only occur every hundred years.  It means that there is a one in one hundred chance of experiencing that flood each year.  Every flood is independent of each prior flood.  A flood does not know that you just had a hundred year flood and it shouldn’t flood you for another hundred years. That is not how probability works.  Each event may be, and probably is, independent from previous events. 

https://www.youtube.com/watch?v=gOwLEVQGbrM

Knowing that  sh*t will hit the fan, means that you better have a plan for when the sh*t hits the fan, which better NOT be that the sh*t will never hit my fan.  That is why good engineering balances efficiency with resiliency.  It does NOT try to increase efficiency by decreasing resiliency.  Resiliency is figuring how likely it is that the sh*t will it the fan, what are the consequences when the sh*t hits the fan, and what to do when the sh*t hits the fan.  Counting on the sh* t never hitting my fan and maximizing my system might make for have an efficient system, but it will not make for a very resilient system.

When you board an elevator there is typically a weight limit posted.  It does not mean that if you exceed that weight limit by one ounce, then the elevator will fail.  That limit lets you know that you are experiencing an unacceptable risk if you exceed that amount.  This might be called a safety margin, a design standard, or…. resiliency.

In queuing there is an amount where the queue starts rapidity going to infinity.  This generally happens if the arrival volume exceeds 80% percent of the service volume, capacity.  If you exceed this point, then if the system fails, you will have an extraordinary problem recovering from that failure. In traffic engineering this point is typically Level of Service “C” or “D”.  In rail operations it is called a parametric capacity, which is also about 80% of the physical capacity.  It might seem very efficient to be at 100% of capacity, but that is NOT a very good idea.  That is why there are guardrails, design standards, safety systems, etc., so that you don’t sacrifice efficiency for resiliency. That is also why well engineered systems have redundancy, so that when one item fails, another item can pick up the load.  It is also why females have two mammary glands even though when they only have one offspring.  Nature is resilient, not just efficient. Let’s learn from nature.

Thinking about sh*t hitting the fan is not very pleasant. Why do you think that insurance companies have cute mascots like geckos, emus, Snoopy or ducks. My favorite is the AFLAC Duck getting confused by Yogi Berra. “When you’re hurt and miss work, it doesn’t hurt to miss work”,  “Huh?”  Don’t be confused yourself.  Efficiency at the expense of resiliency is bad.

No comments:

Post a Comment