Note that the current notation does not include the implementation level. HW
(2S) just indicates duplication, but not whether the whole system is duplicated or
just parts of it, such as, for example, duplicated memory.
Obviously, any redundancy (hardware and software) needs additional structural
redundancy for its implementation. Instruction repetition HW(nT), for example,
needs additional hardware registers to store the internal state for the instruction
rollback.
We call this redundancy supportive redundancy, i.e., redundancy needed for the
implementation of the main redundancy technique. For the sake of clarity, we
usually omit this supportive redundancy. In cases where it is not clear whether an
applied redundancy type is supportive or not, more than one redundancy type can
be used.
Software-based hardware checks that are performed in idle time of the system
are such an example, indicated by SW(∂S, ∂T).
Tables 3.2 and 3.3 show some concrete examples of hardware-, and
software-based redundancy. The reader, using own experience, can extend this list
of examples.
Note that the fault tolerance of a computer system is not considered as a system
feature. Throughout the whole work, we will consider it as a process.
Table 3.2 Examples of hardware-based redundancy
Redundancy
type
Description
HW(2S)
Structural (material) redundancy of hardware such as a duplicated memory
subsystem
HW(S 1 , S 2 )
A duplicated fault-tolerant computer system with principally nonidentical
parts
HW(dI)
Redundant information bit, for example, an additional parity bit per data
word in hardware memory for error detection
HW(nT)
Special hardware to repeat or delay computing to avoid the influence of a
malfunction
HW(dT)
Special hardware to delay execution (like in a timing diagram) to avoid
malfunctions
Table 3.3 Examples of hardware-based redundancy
Redundancy
type
Description
SW(2T)
Double repetition of the same program on the same hardware, generally used
to check the correctness of the results
SW(dI)
Informational redundancy of the program, for instance, a backup file, which
can be used to restart and recover the program state
SW(S 1 , S 2 )
Two different programs, realizing the same functionality in diverse ways. For
example, N-version programming [6]
3.2 Connection Between Reliability and Fault Tolerance
17
(2S) just indicates duplication, but not whether the whole system is duplicated or
just parts of it, such as, for example, duplicated memory.
Obviously, any redundancy (hardware and software) needs additional structural
redundancy for its implementation. Instruction repetition HW(nT), for example,
needs additional hardware registers to store the internal state for the instruction
rollback.
We call this redundancy supportive redundancy, i.e., redundancy needed for the
implementation of the main redundancy technique. For the sake of clarity, we
usually omit this supportive redundancy. In cases where it is not clear whether an
applied redundancy type is supportive or not, more than one redundancy type can
be used.
Software-based hardware checks that are performed in idle time of the system
are such an example, indicated by SW(∂S, ∂T).
Tables 3.2 and 3.3 show some concrete examples of hardware-, and
software-based redundancy. The reader, using own experience, can extend this list
of examples.
Note that the fault tolerance of a computer system is not considered as a system
feature. Throughout the whole work, we will consider it as a process.
Table 3.2 Examples of hardware-based redundancy
Redundancy
type
Description
HW(2S)
Structural (material) redundancy of hardware such as a duplicated memory
subsystem
HW(S 1 , S 2 )
A duplicated fault-tolerant computer system with principally nonidentical
parts
HW(dI)
Redundant information bit, for example, an additional parity bit per data
word in hardware memory for error detection
HW(nT)
Special hardware to repeat or delay computing to avoid the influence of a
malfunction
HW(dT)
Special hardware to delay execution (like in a timing diagram) to avoid
malfunctions
Table 3.3 Examples of hardware-based redundancy
Redundancy
type
Description
SW(2T)
Double repetition of the same program on the same hardware, generally used
to check the correctness of the results
SW(dI)
Informational redundancy of the program, for instance, a backup file, which
can be used to restart and recover the program state
SW(S 1 , S 2 )
Two different programs, realizing the same functionality in diverse ways. For
example, N-version programming [6]
3.2 Connection Between Reliability and Fault Tolerance
17