Redundancy is a term used to signify that in case something fails then there is a backup system available to take up the functions immediately. Redundancy has various configurations. From 1 backup for 1 unit to 1 backup for n number of units. Plus there is also the concept of how the redundancy is achieved and so on.
While fault tolerant simply means that the system can tolerate certain faults. For. Eg. Short circuit and so on.
A redundant system may not necessarily be a fault tolerant system. In case the individual unit say an analog card is incapable of handling the fault f short circuit, and there is a backup card, so when the changeover occurs, then it is possible that the second card may also fail.
Answer: Fault Tolerance includes a standby/back up operation/server/router which can be warmed on loss of primary. The system tolerates primary faults.
Redundancy implies that parallel operation/server/router exist so that if one fails, then the redundant operation can carry the load until the parallel is reinstated.
Though redundancy implies fault tolerance, a fault tolerant system does not have to be redundant.
Fault tolerance simply means that a system can tolerate faults.
Every system can tolerate certain faults for eg. some may tolerate short circuits, voltage fluctuations (within limits) and so on. Fault tolerance has limits (u can see the specifications for knowing the faults that a system can tolerate). Say for theortical example, a system can tolerate 5000 V overvoltage. this means that the system cannot tolerate 10000 V.
Redundancy simply means that if one system fails, then another will take over its functions. Say for eg. a redundant plc has two processors, one main and one standby, then in the event of failure of the main processor, the standby processor takes over.
A redundant system is generally fault tolerant to system faults and faults within the specified limits provided proper engineering practices are
followed and the suppliers recommendations are adhered to and maintained.
In practice many redundant systems are termed as fault tolerant as the system takes into consideration the most common faults enountered in their respective industey, like over voltage, short circuit, failure of processors, failure of communication modules and so on. The suppliers also give guidelines for engineering (say standard wiring diagrams and so on).
However, the fault tolerance of a system ( say a process control system) is dependent not only on the redundant architecture but also on good
engineering, implementation and maintenance.
For theorotical eg. Suppose you have purchased a 1:1 redundant PLC with say even analog I/O redundancy, but if your cable is unshielded and running close to a motor operating infrequently then u may get noise in both the analog inputs whenever this motor runs and it is possible that your plant may trip or operate unsafely or face major disturbance.
Please note that redundancy does not imply that your control system is fault tolerant. You need to understand the limitations of the redundant system and follow correct engineering practices and may have to consider additional factors before your control system can become fault tolerant (again faults within limits).
Fault-tolerant is usually reserved for triplicated systems. They can tolerate a single safe or dangerous failure and continue to operate
properly. Redundant can either mean dual or triplicated. There are many ways of making dual systems.
Responding to Paul Gruhn's reply to A. Milad (Fri, Aug 16, 1:20 pm):
Your position that Fault-Tolerance " is usually reserved" for TMR is biased. It is however, a major marketing tool... and those who promoted
it are to be lauded for its (marketing) success.
Fault-tolerance was the norm for my company's solid-state approach to SIS, regardless of voting hierarchy selection. And, of course, you are
well aware of my thesis... Input/Output diagnostics is the key factor in determining performance, not how many computers are used! I find it strange that the embraced technology (multiple computers) is employed to cover for the unreliability of its principle SIS component, as well as its attendant "errors and omissions" liability! The latter, alone, will insure the demise of smaller vendors.
In closing "to the victor go the spoils" and larger profits.
Facebook
Google
GitHub
Linkedin
