Things that cannot go wrong

Found this Douglas Adams quote in The Signal and the Noise (a must-read book):

The major difference between a thing that might go wrong and a thing that cannot possibly go wrong is that when a thing that cannot possibly go wrong goes wrong it usually turns out to be impossible to get at or repair

I’ll leave to your imagination how this relates to stretched VLANs, ACI, NSX, VSAN, SD-WAN and a few other technologies.


  1. Can I add LISP to that list? :)
  2. SD-WAN Vendors, "Hey its proprietary, don't worry software bugs never happen in our code"

  3. I'm not sure that the technologies you mention are any better or worse than the old ones (except stretched VLAN's, I'll give you that one). Traditional technologies can and do still fail in new and exciting ways, so that reads a little bit like "new technologies are worse than old ones".
    1. In this particular case I'm not judging the quality of the technologies (apart from stretched VLANs), but the complexity behind the scenes, and the impossibility of fixing something when it's broken. Even in the stretched VLAN case, when it's broken, it's often impossible to fix because bridging.
    2. If the software is broken then how is waiting for a fixed version of IOS different to waiting for a fixed version of some vendors SD-WAN code? or VSAN code?

      Personally, I don't believe that new technologies are somehow more complex than the old ones, we've had massive complexity all over for a long time.

      Saying that, the complexity topic is an interesting one though - how do you measure that? How do you decide that one solution is more complex than another one, or that any particular solution is "too complex"?

    3. Have you ever tried to troubleshoot one of the distributed systems that hide all their complexity behind a seemingly simple GUI (or API)?
    4. Yes, not much is new after 30 years in IT :-)

      But isn't that a worthy goal? I mean hiding complexity behind a simple GUI (or API)? If we can't reduce complexity, is the answer to always have a complex GUI (or API)?
  4. Nate Silver's book and theories are good, so long as you're not trying to predict the outcome of an election. ;)
Add comment