A component in a computer system which will, if it fails, stop the system from working. Building
high availability clusters mostly consists of making sure you have no single point of failure, though sometimes some will be tolerated - e.g. by having only a single
passive backplane for a system, on the assumption that this has such a low likelihood of failing that it is a non-problem.
A system with no SPOFs can still be vulnerable to catastrophic failure.