Performance Testing | 35
GUI SCREENS ARE LAID OUT TO LEAD YOU THROUGH EXTRACTING TWO PIECES OF RANDOM, YET RELATED
ALL THAT IS REQUIRED TO MAKE THE TEST CORRECT DATA FROM AN APPLICATION RESPONSE WITH NO SCRIPTING!
because the tools are complex, and highly public performance failure to
you are asking testers to effectively assist them in ensuring this will never
become developers (of test scripts). occur again. When analysing the cause
If anyone asked us in another context for the failure, it usually becomes
if we agreed that developers and apparent that the tests executed
testers had different mindsets and contained one or more of the following
abilities, would we disagree? Yet, we ‘defects’:
still ask our performance testers to • Not (realistic) enough variation of
write programs for 70 percent of their data supplied to the application;
time! • Not enough variation of routes
If you have not had the extensive through transactions (users rarely do
training, and gained considerable A, B, C …);
experience using traditional test tools, • Not enough coverage of business
you will forever be asking yourself transactions;
“what is the function that I need to do • Not enough simulated users.
this?” ‘Fighting’ the tool will distract For those of you that test web
you from ensuring your test scenarios applications, ask yourself this question:
are realistic. it is well know that users leave a web
site at all stages of a transaction; do
Performance testing – your test scripts have a percentage
the reality of users continuing no further after
Some organisations are lucky enough every page? If not, you are not truly
to have the project time, the resources, mirroring real world usage, and
and the tool experts to plan, configure therefore your test results could be
and execute realistic performance erroneous (users leaving a site often
tests. They can test as needed, and still have sessions on the server
release their applications with a high consuming resources).
degree of confidence that they will When you think about it, it is
perform and scale as required. amazing that even more systems
However, it is almost a weekly do not fail. It would not surprise me
occurrence that a major organisation’s if we were to see more and more
high-profile application fails due to application performance failures as
the load applied by real users. This is IT organisations decide there is no
particularly true on the web where alternative to what they currently do
every outage is reported extensively. and simply cut corners.
As it is reasonable to assume that
these applications have all been What is the alternative?
When you think about it, it is amazing
formally performance tested, why do Whether it is because of previous that even more systems do not fail.
these failures still keep occurring? damaging application performance
It would not surprise me if we were
The obvious answer is that the tests problems, monetary pressures, or
to see more and more application
executed were not representative of lack of tool expert skills, increasingly
what actually happened when the many organisations are adopting a
performance failures as IT organisations
application was given to real users. new approach based around different decide there is no alternative to what they
I have been called into a number of performance testing process and tools.
currently do and simply cut corners.
organisations in the aftermath of a The key process change is to mimic
T.E.S.T | June 09 June 09 | T.E.S.T
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52