When looking to alleviate single points of failure, why is it not feasible to create two copies of everything?

What will be an ideal response?

you can't simply make two of everything. That would create far too much unnecessary hardware and administration. Sure, redundancy is fairly easy to do, but the trick is to determine where the redundancy is needed to avoid single points of failure without too much complexity, cost, or administration. We do this process by identifying two things: critical assets and critical nodes.