1 Reply Latest reply: Feb 15, 2012 4:05 PM by Clinton Jones RSS

Recording sapgui activity in HP LoadRunner

Joan B. Altadill
Currently Being Moderated

Hi,

 

We are planning to do some stress tests in our system which at peak time has about 2.000 users logged on the system. We plan to use HP LoadRunner but as far I know there is no way to automatically record user activity, and we will need to record user activity transaction by transaction.

 

Is it true ? What do you recommend ?

 

Regards,

Joan B. Altadill

  • Re: Recording sapgui activity in HP LoadRunner
    Clinton Jones
    Currently Being Moderated

    You will need to choose the scenarios that you want to use to do the stress test.

    An example might be to create a sales order with VA01. You will need to do a recording of the SAPGUI VA01 transaction and then when you load your test bench, specify the number of threads of VA01 that you want to execute concurrently.

    It would likely be better to choose a VA01 with a dozen or so line items and perhaps some line item scheduling in order to get the benefits of a deeper UI experience.

     

    An optimal testing scenario is one where you mix up the testing scenarios, so for this I would recommend examining your key business processes and in particular the ones that are most crucial to business continuity and at the same time are representative of what those 2000 users will be doing.

     

    Accordingly you might find that you choose 5 - 10 scenarios and run these all in parallel. Say order entry with VA01, Purchase Reqs with ME51, Journals with FB50, Material Creation with MM01 etc. The number of sessions for each, will largely depend on the matrix that you have defined for your organization.

     

    When used together with the rest of the HP testing suite in combination with Solution Manager you get the advantage of deeper associated statistics and certainly loadrunner will outline shortest and longest running script with averages and when the peaks and troughs were witnessed. If you are not using the whole suite correlation will require addtional manual effort to determine what was happening at a particular time in relation to perhaps a lomng running process thread.

     

    In terms of overall analytics, if you are not using solution manager you will need to have your basis team use a combination of SM50 and then ST03 to review general system performance. Again the recommendation would be that you use CCMS for general monitoring and use this in combination with Solution Manager in order to get centralized trending and then use that as the baseline metric for setting up general system alerts.

     

    Your BASIS resources will know to look at dialog response time, the number of available dialogs and other factors to determine if the system is underspecified for what you are trying to do, or perhaps needs some tuning.

Actions