4.5 Reliability Evaluation for Fault Tolerance
From the classic reliability point of view, any extra redundancy of hardware reduces
the absolute reliability of the system [11]. At the same time, the reliability of the
system might be increased if the introduced redundancy is able to tolerate faults of
the main part of the system by means of GAFT and itself contributes less in the
system fault ratio.
Thus, part of the problem, the decreasing reliability of the system caused by
hardware redundancy at some point becomes part of a solution. GAFT implementation breaks down to three processes: P1 for checking and testing, P2 for
preparation for recovery, and P3 for recovery and recovery monitoring.
Clearly, there are design trade-offs to be made to achieve the optimum operational reliability and redundancy types used as we have already shown.
The reliability analysis [12], introduced in Sect. 4.3.1, shows a reliability value
for each element (hardware redundancy) and assumes a Poisson failure rate k. This
allows to calculate the overall reliability as a function of time for the whole system,
if the structure of the system is known.
The hardware redundancy used at various steps of GAFT degrades in reliability
over time; thus, the achievable performance and reliability and their degradation
within the life cycle of the RT system are dependent on each other.
Therefore, an analysis of the surface shape and evaluation of performance and
reliability degradation caused by the used redundancies should be performed for
every fault-tolerant system. Figure 4.4 presents qualitatively a slope where a
fault-tolerant system should be located, between the plane of requirements and
curves of reliability and performance degradation.
Furthermore, the introduction of the cost to implement each proposed solution
allows to summarize the system overheads required to implement fault tolerance.
There is no doubt that a quantitative evaluation of reliability, performance, and
cost overheads within one framework might be extremely efficient for justification
of the design decisions and comparison of different approaches in implementation
of fault tolerance. There is a correspondence between reliability of FT systems and
steps of GAFT related to the malfunction tolerance shown in Fig. 4.4.
Two dimensions: reliability and performance correspond to requirements of the
system with controlled degradation over time. This figure illustrates that it is not
possible to minimize for cost and maximize all other goals and a trade-off has to be
made.
4.5 Reliability Evaluation for Fault Tolerance
37
From the classic reliability point of view, any extra redundancy of hardware reduces
the absolute reliability of the system [11]. At the same time, the reliability of the
system might be increased if the introduced redundancy is able to tolerate faults of
the main part of the system by means of GAFT and itself contributes less in the
system fault ratio.
Thus, part of the problem, the decreasing reliability of the system caused by
hardware redundancy at some point becomes part of a solution. GAFT implementation breaks down to three processes: P1 for checking and testing, P2 for
preparation for recovery, and P3 for recovery and recovery monitoring.
Clearly, there are design trade-offs to be made to achieve the optimum operational reliability and redundancy types used as we have already shown.
The reliability analysis [12], introduced in Sect. 4.3.1, shows a reliability value
for each element (hardware redundancy) and assumes a Poisson failure rate k. This
allows to calculate the overall reliability as a function of time for the whole system,
if the structure of the system is known.
The hardware redundancy used at various steps of GAFT degrades in reliability
over time; thus, the achievable performance and reliability and their degradation
within the life cycle of the RT system are dependent on each other.
Therefore, an analysis of the surface shape and evaluation of performance and
reliability degradation caused by the used redundancies should be performed for
every fault-tolerant system. Figure 4.4 presents qualitatively a slope where a
fault-tolerant system should be located, between the plane of requirements and
curves of reliability and performance degradation.
Furthermore, the introduction of the cost to implement each proposed solution
allows to summarize the system overheads required to implement fault tolerance.
There is no doubt that a quantitative evaluation of reliability, performance, and
cost overheads within one framework might be extremely efficient for justification
of the design decisions and comparison of different approaches in implementation
of fault tolerance. There is a correspondence between reliability of FT systems and
steps of GAFT related to the malfunction tolerance shown in Fig. 4.4.
Two dimensions: reliability and performance correspond to requirements of the
system with controlled degradation over time. This figure illustrates that it is not
possible to minimize for cost and maximize all other goals and a trade-off has to be
made.
4.5 Reliability Evaluation for Fault Tolerance
37