Software system failures

These data give the failure times (in CPU seconds, measured in terms of execution time) of a real-time command and control software system. The data can be used as an example for fitting various reliability models. Musa et al. recommend fitting a nonhomogenous Poisson process with linearly exponentially decreasing failure time.

Usage

software

Format

A data frame with 136 rows and 2 columns:

time

Time to failure in CPU seconds

failure_id

Failure number (1 to 136)

Source

Musa, J.D., Iannino, A. and Okumoto, K. (1987) Software reliability: measurement,prediction, application, New York: McGraw-Hill, 305.

Examples

# Distribution of the failure time
with(
  software,
  hist(
    time,
    main = "Distribution of the failure times",
    xlab = "Failure time (CPU seconds)"
  )
)


# Plot of the cumulative number of events over time
with(
  software,
  plot(
    time,
    failure_id,
    main = "Cumulative number of failures over time",
    xlab = "Time (CPU seconds)", ylab = "Number of failures"
  )
)
# Add lines
with(software, lines(time, failure_id))