December
1999 QUESTION 4 Total Marks: 15 Marks |
Click here to access other
questions
SUGGESTED SOLUTIONS |
(a) |
What is system performance evaluation? System performance evaluation is a tool that is used for the systematic determination of the productivity of an object.
|
[1] |
(b) |
System measurement is one stage in the process of system evaluation. (i) Outline the main activities to be performed during this stage. [2 marks] The purpose of system measurement is to gain system performance data (1 mark). Once information is collected, it must be processed; techniques such as simple statistics and multivariate analysis may be used (1 mark). (ii) Explain the difference between hardware
and software monitors. A hardware monitor is a physical tool that records the electrical signals to a device; a software monitor records software messages instead. (iii) Give one advantage of using each type of monitor to record the activity of a device. [2 marks] An advantage of using a hardware monitor is that it does not degrade system performance (1 mark). On the other hand, a software monitor can collect higher-level information, such as who is using the device (1 mark). (iv) Explain the difference between sample and trace modes of monitoring. [1 mark] In sample mode, monitor remains dormant most of the time, waking up periodically to sample system activity; in trace mode, the monitor remains continuously active.
|
[6] |
(c) |
Outline the four activities that may be carried out during software maintenance. diagnosis and correction of errors; modification of software to interface with changing environments; satisfying requests from users for new facilities, modifications to existing functions, and general enhancements; improving internal structure for future maintainability or reliability.
|
[4] |
(d) |
The diagram below shows the failure curve for software.
(i) Explain how the idealized curve is derived. [2 marks] Software does not suffer from wear and tear like hardware (1 mark); therefore, after the initial phase in which errors are found and eradicated, one would expect the failure curve to remain constant, or even to drop to zero (1 mark). (ii) Why does the actual curve differ from the idealized curve? [2 marks] Throughout the life of software, maintenance is carried out, which is likely to introduce new defects (1 mark); moreover, changes erode internal structure, making errors more likely as time goes by (1 mark). |
[4] |