|
In order to create reliable real time systems, it is important to know not only
what the expected capacity and response time is in normal conditions, but also what
the behavior of the system will be in degraded conditions. In systems which include
redundancy or allow for degraded operations in the presence of non-critical failures,
this assessment must be integrated with the expected reliability and/or availability
of each operational state. Typically, response time and performance evaluations
for complex systems are performed using discrete event simulations. Reliability
and availability predictions can also be done using discrete event simulations or
analytical methods. However, the time scales of response times and failure rates
can vary by 12 orders of magnitude, well beyond the feasible range of any simulation
program to perform. We have developed a method to evaluate integrated reliability
and performance using standard reliability and performance measurement in an innovative
technique that combines both Markov and discrete event simulation methods.
|
|