cancel
Showing results for 
Search instead for 
Did you mean: 

Why does HP LoadRunner measure slow response time for SAP GUI selection steps?

0 Kudos

We are running HP LoadRunner 11.52 scenarios out to multiple (up to 4) load generators.  Load generators have 24 GB ram and 8 CPU's.  We have run a maximum of 200 users and see a maximum usage of ~25% on CPU's.  We were previously running SAP GUI 7.4 and downgraded to 7.3 when we came to understand 7.4 is not supported in PAM for LR 11.52.  Regardless, our results have not changed.  What we are trying to resolve is why the first two steps of opening the SAP GUI and entering system-related selection is taking so long to process?  Additionally, we typically see that during the time escalation of those steps, that there is a period where the process seems to pause for a large time gap.  The process below shows a snapshot while in progress.  The SAP GUI steps are hitting severe peaks wthile all other steps during the process tend to track at a steady quick response time.  Any degradation there looks to be tied to the long pauses.

Would anyone have knowledge or experience with this type of output and understand why it might be consistenty happening?  Thanks.

Accepted Solutions (0)

Answers (2)

Answers (2)

Former Member
0 Kudos

Mark,

Please can you let me know who you are and what industry you organisation are involved in.   I am seeing a similar issue and work in a very large organisation and am just checking that these are separate instances of the problem.

Appreciated.

0 Kudos

Hi,

My suspicion is that SAP GUI start-up just requires the associated amount of resource.  After having been live for 6 months, I believe now that's normal as it's opened / closed / opened, etc.  Haven't had any response time concerns and we are typically running close to 200 concurrent sessions.

Former Member
0 Kudos

Mark,  thanks we are seeing similar issues, where the LGs are v11.x, however where the LGs are v12.x these issues do not exists.

You have done something similar, where you have upgraded to 7.4 and downgraded due to support, this aligns also.

We have wild response times, but it does stabilise a little once all the users are on board.

Can I clarify that you are running 50 vUsers as a process, per LG.  Also are these physical or virtual systems you are using.

Hoping that we might get to the bottom of this if we can shared knowledge.

msahukar
Explorer
0 Kudos

Hi Mark,

Did you ever get an answer about this issue? We are experiencing the same situation and would like to know the cause.

Thank you!

0 Kudos

Never got an answer.  The only way we could dramatically decrease the total time for each event was to distribute the processes to additional Load Generator machines.  In other words, the pattern was the same regardless.  Peak times however decreased marginally with each LG added - nearly 50% each time - into an acceptable range.  Don't know enough about LR to say that it's a common demonstration of opening the GUI.   Would still like to know bottleneck cause.  

Former Member
0 Kudos

Maria,

Same question as posed to Mark, its the pause in the ramp-up that grabs my attention so just want to check that we are not the same organisation.

Appreciated.