In basic redundancy, just N identical supplies are needed for the load, but N+1 are used in parallel, where N can be as low as one but is typically a number between two and six. If any one of the N units fails the remaining units can provide the full, needed output current. This approach works because the probabilities of more than one unit failing are quite low.  For example, if the reliability of any single unit is 0.99, then the probability of two units failing is 0.9999 in an N=1 design. The N+1 method does bring higher up-front costs, of course, and often hot-swap capability to replace the failed supply in needed.