Why standard.txt and chartrand.txt used in jmeter ?

[nikhilgarg]

Hey ,  I am using JMeter with Qlikview for stress testing but can anyone tell me why is standard.txt and chartrand.txt used in that ??  Thanks

[Sebastian_Fredenberg]

Hi Nikhil,

Those files are used as input seednumbers for randomizing selections (if you have selected seeded randomization).

/sebastian

[nikhilgarg]

Hey

Thanx for telling but i really didn't understand . Wat is seeded

randomisation ? Help me plz

On 27-Jan-2015 7:18 PM, "Sebastian Fredenberg" <qcwebmaster@qlikview.com>

[Daniel_Larsson]

It means that the randomized sequence will be the same each time the sequence is created.

In this case if you run the exact same script twice, the selections made by each user will be the same in both runs. I.e. User A will select value X in object Y both script executions. If you choose completely random User A will select value X the first run and value W the second run. With a completely random selection the two execution would not be comparable since the calculations triggered are not the same.

[nikhilgarg]

Hey thanx a lot daniel.

So wat i understand from this is:

Let say i have created a script that records accessing access point,

selecting a dash board, selecting a value say "abc" from list box.

Now this recorded script i want to execute for 10 virtual users. But all

the virtual users will now select value "abc" from listbox. So to avoid

this and we want each of 10 virtual users should select different value

from listbox, we should use standardrand.txt and chartrand.txt. Is it so??

Have i understood it ryt ??

[Sebastian_Fredenberg]

Hi again,

Using the files means that each time you execute the script, the same values will be selected in the same sequence as other times - but it does not mean that each user will select the same values as other users within one execution.

As an example, if you have execute the script described above and do not use seeded randomization you might get the following selections from that listbox:

Execution 1: User 1 - Selection A, User 2 - selection B, User 3 - C, User 4 - D, User 5 - E.

Execution 2: User 1 - B, User 2- D, User 3 - E, User 4 - A, User 5 - B.

So you have no control over what will be selected each time the script is executed, and can make it difficult to compare results between 2 tests.

Whereas using seeded randomization it will be more like this:

Execution 1: User 1 - A, User 2- B, User 3 - C, User 4 - D, User 5 - F.

Execution 2: User 1 - A, User 2- B, User 3 - C, User 4 - D, User 5 - F.

Which means that you will still randomize selections between the users, but each time you execute the script the same selections will be performed. That means that you can tune the script, app or environment between tests and have a trustworthy comparison.

/sebastian

[nikhilgarg]

Hey Sebastian,

Thnaks it really helps . But this is first time i am doing Stress testing so can you tell me how after including standardrand.txt and chartrand.txt , seeded randomisation works ??

THanks

[Sebastian_Fredenberg]

Hi,

Actually you don't need to have a deeper knowledge of how the seeding works in order to perform the stress test, just that if you select the "from file" option when creating the script in the GUI (default option) then the same random selections will be performed every time the script is executed. You do not need to do anything with the files themselves, just leave them at where they are after unzipping the scalability tools package.

But if you want to know how it works I can try to explain. The files are populated with pregenerated values that is used as input for the selections. These values are not the actual values that will be selected by the virtual users, instead they represents relative values/positions to select. When the server sends a response with values to select from, the actual selection is picked using the position value as input. That position/seed could lead to value x is picked from one response or value y from another response. As the same files are used every time a script is executed, the same positions will be used from the responses which leads to the same selections being performed every time for a specific object+user. So the selections to be performed will be random, but the randomization is controlled in the way that you can rerun a test and simulating the same behaviour over and over.

If you are not interrested in having this control, then you can uncheck the "from file" checkboxes when creating the script - which means that randomization will be performed at runtime instead without the predicting what the values to select will be.

/sebastian

[nikhilgarg]

HEy thanx . I am stuck in polling request. In jmeter it is said to add :

Xpath extractor poll objects and Poll string Builder Objects. But i cann't find this in jmeter.

Cna you help em in this ??

[Sebastian_Fredenberg]

Hi,

I believe you are using an old way of creating the scripts, as in the current scalability tools package you do not need to do anything with those components (there is a tool that builds the script for you).

Please check here for the current version of the tool, as it will simplify everything

QV Scalability Tools (QV10,QV11)

/sebastian