CSE 322: Software Reliability Engineering Topics covered: Architecture-based reliability analysis

Failure models of components  Three types of failure models  Probability of failure or reliability:  Constant failure rate:  Time-dependent failure intensity:

Steps in architecture-based analysis Architecture-based models Architecture of the application Failure behavior of components Solve the model Superimpose + Performance predictions Performance bottlenecks Reliability predictions Reliability bottlenecks Description Solution

Combinations of architecture & failure behavior Absorbing DTMC Probability of failure, R Constant fail. Rate, Time dep. fail int. ( t ) Irreducible DTMC Absorbing CTMC Irreducible CTMC Model #1 Model #2 Model #3 Model #4 Model #7 Model #6 Model #5 Model #8 Model #9 Model #10 Failure behavior of components Architecture Consider only DTMC and CTMC based models Shaded area indicates solution infeasible using DTMC and CTMC models

Model #1

Example - Model #1 Terminating application 10 modules 1 is the input, 10 is the exit module Architecture modeled by absorbing DTMC Reliability, mean execution time of each component known Probability that the control is transferred to module j, upon execution of module i Reliability of component i Execution time of component i per visit - Time to completion of the application - Reliability of the application

Example - Model #1 (contd..) Component Reliability Mean execution time per visit Known for component i Reliability Time to completion units Computed overall

Combinations of architecture & failure behavior Absorbing DTMC Probability of failure, R Constant fail. Rate, Time dep. fail int. ( t ) Irreducible DTMC Absorbing CTMC Irreducible CTMC Model #1 Model #2 Model #3 Model #4 Model #7 Model #6 Model #5 Model #8 Model #9 Model #10 Failure behavior of components Architecture Consider only DTMC and CTMC based models Shaded area indicates solution infeasible using DTMC and CTMC models

Model #2

Example - Model #2,Terminating application, 10 modules 1 is the input, 10 is the exit module Architecture modeled by absorbing DTMC Time independent failure rate known Execution time of the component per visit known Probability that the control is transferred to module j, upon execution of module i Mean execution time of component i per visit Failure rate of component i - Time to completion of the application - Reliability of the application

Example - Model #2 (contd..) Component Failure rate Known for component i Expected execution time of the component per visit Reliability Time to completion units Computed overall

Combinations of architecture & failure behavior Absorbing DTMC Probability of failure, R Constant fail. Rate, Time dep. fail int. ( t ) Irreducible DTMC Absorbing CTMC Irreducible CTMC Model #1 Model #2 Model #3 Model #4 Model #7 Model #6 Model #5 Model #8 Model #9 Model #10 Failure behavior of components Architecture Consider only DTMC and CTMC based models Shaded area indicates solution infeasible using DTMC and CTMC models

Model #3

Example - Model #3,Terminating application, 10 modules 1 is the input, 10 is the exit module Architecture modeled by absorbing DTMC Time dependent failure rate and mean execution time of each component per visit known Probability that the control is transferred to module j, upon execution of module i Time spent in component i per visit Failure rate of component i - Time to completion of the application - Reliability of the application

Example - Model #3 (contd..) Component Failure rate Known for component i Expected total number of faults Mean execution time of the component per visit Fault detection rate per fault Failure rate of the Goel-Okumoto model Reliability Time to completion units Computed overall

Combinations of architecture & failure behavior Absorbing DTMC Probability of failure, R Constant fail. Rate, Time dep. fail int. ( t ) Irreducible DTMC Absorbing CTMC Irreducible CTMC Model #1 Model #2 Model #3 Model #4 Model #7 Model #6 Model #5 Model #8 Model #9 Model #10 Failure behavior of components Architecture Consider only DTMC and CTMC based models Shaded area indicates solution infeasible using DTMC and CTMC models

Model #4

Example - Model #4 Non-terminating application 10 modules Architecture modeled by irreducible DTMC Reliability of each component known Probability that the control is transferred to module j, upon execution of module i Reliability of component i - Utilization of module i - Reliability of the application

Example - Model #4 (contd..) Component Utilization of the component Reliability of the application Computed for component i Computed overall Known for component i Reliability of the component

Combinations of architecture & failure behavior Absorbing DTMC Probability of failure, R Constant fail. Rate, Time dep. fail int. ( t ) Irreducible DTMC Absorbing CTMC Irreducible CTMC Model #1 Model #2 Model #3 Model #4 Model #7 Model #6 Model #5 Model #8 Model #9 Model #10 Failure behavior of components Architecture Consider only DTMC and CTMC based models Shaded area indicates solution infeasible using DTMC and CTMC models

Model #5

Example - Model #6,Terminating application, 10 modules 1 is the input, 10 is the exit module Architecture modeled by absorbing CTMC Reliability of each module known Execution time in each module follows exponential distribution, mean known Probability that the control is transferred to module j, upon execution of module i Mean execution time in component i Reliability of component i - Time to completion of the application - Reliability of the application

Example - Model #5 (contd..) Component Utilization of the component Computed for component i Computed overall Known for component i Failure rate of the component Failure rate of the application