Hi,

Usually the server uses a cookie to keep session for each individual and
unique app/browser that tries to connect. This is not always tied to the a
user's credentials so it is possible to login with a single set of
credentials and test the login realistically. So you just need to make sure
what's the situation with your app.

However, other features of the app can be affected by an use-case such as
this. I.e.: the user's data is in the same section of the database - so
disk and/or ram is/are less used then in production, caching is easier for
the servers so response times are better. And the list goes on. I can't
think of other more concrete stuff right now, but there are a LOT of
possibilities. This is not the case too often and is very difficult to
catch as a tester - so you should simply avoid a monotonous use case as
this if possible and easy for you (in general you should keep an eye for
monotonous input data as a performance tester: once I got into trouble by
using user1, user2, user3... for each new thread which was not a realistic
use-case for that particular app; all my requests were going to a single
machine in the cluster and I was investigating load-balancer issues when it
was the test at fault, it took me and devs a brainstorming meeting to catch
this bugger, after weeks of bad results).

But this is just the default advice, you really need to pay attention to
the architecture of the system and performance indicators (like in my case,
requests were reaching a single machine, CPU and network were highly used
only on that machine). You can use this information to workout those faults
that are mission-critical. Even better is to have a very good idea of
real-users' workload and replicate that as best as possible, but in my
experience this kind of data is often missing - and when this is not the
case, my advise is to just have a large set of performance indicators that
you keep on eye on during tests.

--Adrian S



On Thu, Sep 6, 2012 at 1:49 PM, Divya Singh <[email protected]> wrote:

> Hi All,
>
> I am executing a Jmeter test plan with 30 virtual users and using a single
> login ID for all threads. I also have added a CookieManger to my test plan.
> The e2e transaction I am executing is
> 1) Login to application
> 2) Conduct a Search for a Ticket number
> 3) view the ticket detail
> 4) update the Ticket and save
> 5) Run a report
> 6) Logout
>
> This is a basic test however I have a doubt if it is advisable to use the
> same login credentials for all the simulated threads? Are there any
> implications of using this approach?
>
> Per my understanding, Jmeter simply sends http requests to the server and
> for the server each incoming request is unique. So it should not matter if
> the threads are using the same login.
>
> Please correct me if I am wrong.
>
> Thanks,
> D
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>
>

Reply via email to