Prev | Current Page 128 | Next

Peter Farrell-Vinay

"Manage Software Testing"

17 and the alerts.
3.1.1 Risk Alerts
These are the triggers used to identify when a risk is imminent. Typical test-related triggers are:
??? Reduction in the number of lines of code per bug found.
??? Finding an unacceptably-high number of priority-1 and -2 bugs in a build.
??? Finding an unacceptably-high number of bugs in a component.
??? Late arrival of signed-off specifications for use as a baseline.
??? Failure of performance tests to achieve targets.
??? Growing code complexity.
??? Growing code turmoil.
Monitoring such risks is easier when an alerting system is in place.
The existence of a risk log allows the test team to identify priorities and provides a good basis for
deciding the mix of tests to be planned.
3.1.2 Risks as Drivers
Risks can be grouped by
sources
(as shown above) and by
kinds
. A risk
kind
is for example that something
doesn??™t work, that it works too late, too slowly, at the wrong time, or that it has unintended side-effects.
These groups are sensitive to risk drivers in that a driver can change a whole group of risks. Thus the
product risk
kind
may be composed of the following subrisks:
??? That features A, B, and D don??™t work correctly.
??? That features C, and D are too slow.
??? That features B, and E are highly susceptible to environmental change, and that the proposed
move from database X to Y will cause many problems.
These risks might be related to (say) the failure of the project to use an appropriate
development method
(the project is mostly staffed by new graduates with some experience in Visual Basic, and none in design.


Pages:
116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140