Information Technology Reference
In-Depth Information
as user response times, screen navigations, or the number of
mouse-clicks needed to fulfill a task. By our definition, a critical
requirement
be skipped in development for the software to
be considered a failure. Such a cause of failure points to a basic
mismatch of the perceived criticality of needs between the customer
and the project team.
must
The software was built to specifications but has become unreliable
to a point that its use is being impacted.
Such failures are typical
when software is first released, or a new build hits the QA team
(or the customer). There could be many small bugs that hinder
the user from performing his or her task. Often, developers and
the QA team tend to overlook inconsistencies in the GUI (graphical
user interface) or in the presentation of data on screens. Instances
of these inconsistencies include, but are not limited to the place-
ment of “save” and “cancel” buttons, format of error messages,
representation of entities across screens (e.g., “first name” followed
by “last name” in one screen, and “last name, first name” in
another). These cannot be considered bugs; rather, they are issues
with a nuisance value in the minds of users, “putting them off” to
a point that they no longer consider the software viable.
Failure Patterns
Failure as a field has been studied extensively ever since machines and
mechanical devices were created. In the early days, most things were
fixed when they were broken. It was evident that things would fail at
some point in their lifetime, and there had to be people standing by who
could repair them. In the 1930s, more studies were conducted, and many
eminent scientists (including Waloddi Weibull) proposed probabilistic dis-
tribution curves for failure rates. The most commonly accepted is the
“bathtub curve,” so called because its shape resembles a Victorian bathtub
(Figure 1.1).
The first section of the curve is called the “infant mortality.” It covers
the period just after the delivery of a new product and includes early
bugs and defects. It is characterized by an increasing and then a declining
failure rate. The second section, the flatbed of the bathtub, covers the
usual failures that happen randomly in any product or its constituents. Its
rate is almost constant. The third section is called “wear-out,” in which
the failure rate increases as the product reaches the end of its useful life.
In physical terms, parts of a mechanistic system start to fail with wear
and tear.
