How to calculate the concurrent users for Load Test

Hi,

 

Is there any formula to calculate the number of concurrent users for Load Test?

 

Suppose we have 1000 users in application but we are interested in concurrent users. So what should be the user load that we can achive in LoadRunner scenario

 

Thank you !!

 

Prabhjot

  • I'm not sure i understand your question. Are you asking what is the number of vusers that you need in order to perform an adequate load test of your application? If so, the answer is up to you and your team. You should determine what load you expect (both in regular and peak periods) and allocate the vusers accordingly.

  • Verified Answer

    Hi,

     

    Firstly try to understand your application behaviour, its expected usage, most frequently used transactions or business flows.

    Then try to build a Load Model based on the above information.

    Load Model is nothing but how many VUsers need to be generated to perform X number of transactions over a period of time say 1 hour with a certain think time between each sub transactions and pacing between different scripts.

    Let me explain the User Load in more generic terms
    Named Users:

    All those users who have access to the application. They may or may not log in to the application frequently

    Active Users:

    All those users who are regularly accessing the application day in and day out. They login wait for some time(ideally a few seconds to minutes) and perform some actions and then go to next transaction or business flow. This is the Real User behaviour. The pacing and thinktime for such users can be set based on how many transactions are to be targetted in say 8-9 hours of a business day or based on the usage of the application.

    Concurrent Users or to be more specific Connections :

    These are types of VUsers who keep on performing some activity or the other at any given point of time on the server.

    These can be captured by the active connections on the app/web layer and/or the Database depending on the architecture of the application.

     

    Gramatically speaking we really use Active Users in all our Load, Stress, Endurance Tests and NOT Concurrent users 

     

    To give more clarity, say we have 100 (Active) VUsers performing some activity with some pacing and think time.

    If we drill down to the no of connections at App/Web then we might end up seeing 30 or 40 that too strictly based on Pacing and ThinkTime

     

    Concurrency is inversely proprional to the Thinktime and Pacing. More the think time and pacing, lesser the concurrency and so on and so forth