Scripting efforts in performance testing
Source : SQA Forums : http://www.sqaforums.com/showflat.php?Number=583987
am looking at understanding what is effort in hours that goes in creating scripts.
I usually classify as simple medium and complex and allot around 6-8/10-14/18-24hrs of effort based on the application type/Protocol Used/Number of screens. ( this also depends on the scripter ).
Would like receive inputs on how experts do estimation for creating scripts/or experiences in effort required while creating performance testing scripts.
And I am not limiting myself to loadrunner, I am looking at most of the industry standard tools like SilkPerformer, Rational Performance Tester/WebLoad/JMeter/Grinder.
I usually classify as simple medium and complex and allot around 6-8/10-14/18-24hrs of effort based on the application type/Protocol Used/Number of screens. ( this also depends on the scripter ).
Would like receive inputs on how experts do estimation for creating scripts/or experiences in effort required while creating performance testing scripts.
And I am not limiting myself to loadrunner, I am looking at most of the industry standard tools like SilkPerformer, Rational Performance Tester/WebLoad/JMeter/Grinder.
==============================================================
(Number of Steps * Skew for Protocol * Skew for level of expertise for the scripter with protocol * skew for tool used * skew value for scripter expertise with tool) + (anticipated number of custom functions based upon examination of the app and business processes * 4 * Skew for level of expertise for the scripter) = estimated number of hours per business process scripted
So, if you have a new untrained scripter on a new protocol on a new tool with a couple of custom functions that need to be created and this protocol is winsock, then perhaps you wind up with a estimate range of
Assuming tool and protocol expertise
(8(steps) * [2.5 - 5.0] * 1 * 1 * 1) + (2 * 4 * 1) = Range of [28 to 48]
That gives you a generic formula. You will need to come up with your own values for protocol skew, tool skew, person expertise skew and the like.
So, if you have a new untrained scripter on a new protocol on a new tool with a couple of custom functions that need to be created and this protocol is winsock, then perhaps you wind up with a estimate range of
( 8(steps)
* [2.5 - 5.0](protocol Skew)
* [7 - 10](expertise skew-protocol)
* 1(tool skew)
* 2(Expertise with tool skew )
+ ( 2(custom functions)
* 4(hours)
* [7-10](expertise skew-protocol) )
Range of [336 to 880]
Assuming tool and protocol expertise
(8(steps) * [2.5 - 5.0] * 1 * 1 * 1) + (2 * 4 * 1) = Range of [28 to 48]
That gives you a generic formula. You will need to come up with your own values for protocol skew, tool skew, person expertise skew and the like.
For more information on this, click here
Comments
Post a Comment