If you subscribe to ITIL or Visible Operations ideal, you probably believe that 80% of all IT related outages are self-inflicted due to change.
So for 80% of all IT outages, it does make sense to have a strategy where it is cheaper to rebuild(revert the change) than to repair.
But for the other 20%, it does make sense to investigate further. A virtualization strategy where you could redeploy the offending server while saving the old one out of service for investigation seems ideal.
I agree with investigation in principal, but the blog post seems quite sensational and misleading.