6.4) write an equation thus:
Note that reliability of each item except the legacy system has been squared since they will be
encountered both on the outward and return journeys of the transaction.
3. Get the reliability of the components. We can draw a table of the reliabilities over one hour as
shown in Figure 5.6.
FIGURE 5.5 Simple reliability block diagram
Customer??™s
browser
Browser web-server reliability test
Internet
Web server
Firewell
Site reliability test
Database server
Application
objects
Tester??™s machine
Banking
application
reliability
test
Banking
application
Legacy
applications
Legacy
applications
reliability
test
Application
reliability test
R transaction R R R R sys Br In Fi We ( )( )( )( )( = ??? ??? ??? 2 2 2 ) ( ) ( ) 2 2 ??? ??? R R Ap Le
80 Manage Software Testing
4. Calculate. We can evaluate these figures thus:
which in Excel notation looks like:
which is pretty bad. Assuming that some 500,000 people hit that site every hour some 50,060
(500,000 * (1 ??“ 0.8998792047)) will have a failed transaction. This is entirely due to the lousy
reliability of the web server, the application objects, and the legacy application. If these were all
improved to 0.999999 the reliability would increase to 0.999890005 and only 55 transactions per
hour would fail.
5. Mitigate
a. If we double the web server in parallel, the calculation (with reference to section 18.6.5 in
Chapter 18) is
which when plugged back into the original equation gives us 45,550 transactions failing.
Pages:
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227