6

I want to understand what is the difference for the above mentioned scenario in Load testing.

How different would the results be?

Also another scenario would be wherein I'm having the same user listed 1000 times (users) and executing a scenario.

I am using LR 12.53 and am relatively new. So want to clear my basic understanding.

Kindly help

5 Answers 5

7

1 user doing 1000 iterations is not load testing assuming the user only does one at a time. That said, it is a flavor of performance testing that tests the base product for how long it takes to load the page, access the database, etc. It may need to take into account low bandwidth (i.e. slow network) to see how the application performs under a variey of situations. In some ways this is like testing the load on the services you depend on (i.e. making them slow).

However 1000 users doing one iteration concurrently is the more traditional definition of load testing and is the situation where people typically use a tool such as jMeter. You use the results to tune your application and infrastructure for the real world situation of a large number of concurrent users to be sure your application can scale.

1
  • I don't really disagree in practical terms but just for the sake of discussion... I'd suggest 1 user/1000 tests each could be called "load testing with a very low concurrency and high number of repetitions" and 1000 users/1 test each could be "load testing with a high concurrency and very low number of repetitions".
    – Cherree
    May 21, 2019 at 18:50
2

That depends on what you are testing. assuming you want to test a servers ability to serve static webpages to different users, having only one user would make no sense. The first iteration might give a reasonable result but further iterations might be much faster because of e.g. caching or not having the overhead of establishing a secure connection.

I don't know about loadrunner specifics but i would assume one user only makes request sequentially while 1000 users would send request parallel

2

In most cases, whether LoadRunner is simulating 1 user and having them do 1k actions at the same time or 1k users and having them do just 1 thing, won’t make a difference. Because generally the question of scale is really about number of simultaneous connections. But it can make a difference due to caching and state changes.

Simple example, suppose you want to tell the user something the first time they take an action on a per day basis, suppose that determining what to say takes a long time. If you have 1k different users being used, then the message is definitely going to be composed 1k times. If on the other hand, you have 1 user repeating the same action 1k times, then at least some of them can be expected to skip composing the message, with the corresponding reduction in effort.

2
  • Would it be useful to have 10 simulated users with 100 iterations each so that if load-balancing was in use then it would catch incorrect server-side state cache configuration? May 19, 2019 at 12:48
  • @AndrewMorton: when testing for incorrect server side caching, you don’t need multiple users, one is fine. What you need is multiple iterations. Unfortunately, it’s generally something you have to be looking for under a specific test plan, and not something you can casually add to a load test.
    – jmoreno
    May 19, 2019 at 13:19
2

Let’s say that the action under test takes one second. Then one user doing that 1k times would spread out the test over 1k seconds. 1k users doing it once would try and the same work in one second I.e. at the same time.

So while they do basically the same amount of work, since the time scales are significantly different, then the load is significantly different.

1

I want to understand what is the difference for the above mentioned scenario in Load testing.

How different would the results be?

When 1 user is doing 1000 iterations, it is not load testing. There will not be any load put to the server and most probably server will not face any issues. But when you make 1000 users to do 1 iterations, you can get to know how the server responds under load. You can see a clear variations in response time and server resources utilization, which can be used to draw some conclusions.

Also another scenario would be wherein I'm having the same user listed 1000 times (users) and executing a scenario.

You can do 1000 logins(sessions) with the same user credentials, if the server allows concurrent sessions and depends on the number of concurrent sessions allowed. But it is not advised to have concurrent sessions because it creates extra overhead on security and finally it all comes to business needs.

You will not face any issues from LR side for doing this, you can parameterize the same user name or even hard code the user in the script, since it is 1 user.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.