How do you model user behaviour for performance tests


Performance and scalability: How do you model user behaviour for performance tests

  1. How do you model user behaviour to be input to bots when making performance tests? In our project we have chosen to model everything an average player does under an average session: Average Session length 30 min Login 1 Logout 1 Register 0.1 ... An then calculate the load that one player does on the system and then use that result to scale to X number of players. How do you do it?
  2. Extrapolating cannot be done always.[ Go to top ]

    You are using bots which will gather information about a single site.Fine!Now you are simulating a single user who will just login, do some operations and log out.The analys the performance of the system, you cannot just multiply it with say X users.No the system will behave under high load,there will concurrency issues which will not come up as long as you dont simulate concurrent users(real world scenario).The solution will be 1: Say your bot or your client program simulate one user. Start multiple threads with different login ids and start them together or ramp up, then they will all put load on the application.You will see totally different behaviour, you can see how you application perform under high concurrency. 2: Use products like LoadRunner which can simulate 1000s of users. 3: Talk to me again for more on Load model. cheers,