SoftwareReliability Engineering: Technology for the 1990s

1
Software-Reliability Engineering Technology for
the 1990s

• John D.Musa, William W.Everett, ATT Bell
  Laboratories
• Student Sanaa Kamari

2
What is it?

• Software reliability engineering is the applied
  science of predicting, measuring, and managing
  the reliability of software based systems to
  maximize customer satisfaction.

3
Why is it important?

• Customers of software suppliers have become very
  demanding in their requirements.
• The costs of operational disruptions and recovery
  that the customer encounters due to failures have
  increased.
• Increased demand for quality also increases the
  importance of precisely measuring how well your
  competitors are doing in providing quality to the
  market.
• There appears to be a strong correlation between
  interest in adopting reliability-engineering
  technology and the competitiveness of the
  organization or project concerned.

4
Basic concepts

• Software failure some behavior of the execution
  program that does not meet the customers
  operational requirements.
• Software fault a defect in the code that may
  cause a failure.
• Software-reliability proper is the probability of
  failure free operation for a specified time
  duration.
• Failure intensity refers to the number of
  failures experienced per time period.
• Operational profile the set of functions the
  software can perform with their probabilities of
  occurrence.

5
Setting improvement goals

• Percentage improvement goal in terns of the
  customer-oriented failure vs. the
  developer-oriented fault density.
• Quality-improvement effort on the functions used
  more frequently.
• Example a system performs only two functions
• A 90 of the time. with 50 faults.
• B 10 if the time with 50 faults.
• failure density 0.9(50)k 0.1(50)k 50k
• if you concentrate on the most used function B
  so it has no fault.
• New failure density 0.1(50)k

6
Life cycle

• Software-reliability engineering is applied in
  every phase of the life cycle.
• Definition phase should include the product's
  failure-intensity objective.
• Design and implementation phase should allocate
  the system-reliability objectives among the
  components.
• Validation phase software-reliability
  measurements (Reltools).
• Operation and maintenance phase
  reliability-measurements can help monitor the
  operating softwares reliability.
• Process improvement root-cause-analysis of
  faults.

7
Research opportunities

• Predicting reliability before program execution
  from characteristics of the software product and
  the development process.
• Estimating before test need to find a way to
  estimate the number of remaining faults based on
  patterns of data taken on design errors detected
  on design inspections and coding faults detected
  in desk-checking or code walk-throughs.
• Estimating during test need to develop ways of
  applying software-reliability theories to
  estimating failure during unit test.