StarBase Blog

 

We often talk about levels of concurrency in terms of numbers of users who access a web page or application at the same time. This definition is to not always useful for designing performance tests. To help break this down five different levels of concurrency are discussed below.

 

Level 5

The number of users that could access the application. This is usually the registered users of that system or the total number of users with accounts or provisioned. It is typically unlikely that all of the users will access the system at exactly the same time and as such it is not normally necessary to test this number of users. However this number is needed to help with sizing data for populating the solution and working out later levels of concurrency.

Level 4

The number of users concurrently logged on. This relates to the number of users that are logged onto a system, it does not imply that they are doing anything. Examples include a chat application such as Skype where users may be logged in all day but only actively do something for a few minutes each day. The other extreme is something like an online game where users are active for the whole time they are logged on. This level is useful to know as it allows us to understand the memory footprint we will need to emulate. It can also be the basis for the following levels.

Level 3

The number of users actively using the application. This is the users that are following the steps of a business process. This will typically be the number of users we need to emulate for a performance test. A good example would be the number of users logged into a webmail solution that are reading emails and sending emails. There are inherent delays in actual actions whilst they are reading an email for example. These delays are often simulated with think time delays in the script.

Level 2

The number of users performing an action at exactly the same time. This is the absolute minimum number of Vuser’s that will be required. This can be referred to as Transactions per second as it relates to the number of users who are performing an action at one time, so using the above webmail example it will be the number of users who open an email or reply to an email at any given time.

Level 1

The number of users performing the same action at exactly the same time. This allows us to understand any peaks in business processes we may need to emulate. An example of this is when we do manual performance tests and ask users to press logon at exactly the same time. This is only likely to occur in real life when there are a high number of active users. So for instance if there are 1000 users logged into a webmail solution we may only expect 2 to actually press login at EXACTLY the same time.

 

When planning the number of Vuser’s we need for a performance test we need to consider what the numbers are for all the above levels and come up with a load model which is representative of the highest level possible which is realistically testable with the constraints of cost and time. Need help with your performance testing definitions then consider a performance testing evaluation from StarBase.




 

 

 

Posted by:admin
admin

 Leave a Reply

Your email address will not be published. Required fields are marked *


Security code Cannot read the characters? Refresh

Client Testimonials