A single point of failure (SPOF) is a part of a system that, if it fails, will stop the entire system from working. They are undesirable in any system with a goal of, be it a network, software application or other industrial system. Systems are made robust by adding redundancy in all potential SPOF and is generally achieved in computing through high-availability clusters. Redundancy can be achieved at the internal component level, at the system level (multiple machines), or site level (replication).
The assessment of a potentially single location of failure identifies the critical components of a complex system that would provoke a total systems failure in case of. Highly systems may not rely on any such individual component.