As far as I can think, if you wanted to adapt prep_offset, it should at least
be a function of your network load (bandwidth, latency), of the test systems
themselves, and of the call load.
For the SIPP to be able to make a call at time T, users must be available, in
the right state, and knowing what they have to do. This user preparation must
be done before the call itself. All the parameters above (and others) influence
the time taken this TCP messaging (for instance, when the load increases, the
test systems become less responsive, the network bandwidth increases, and this
preparation time increases).
We have not done any measurement on how low prep_offset can be. When initiating
only one call/second, I guess prep_offset could be much less than a second. But
the question is whether it's worth reducing prep_offset (the test will fail if
the TCP preparation is not completed when the call must start).
We already used some configurations with more than 20000 scenarios/seconds
(multiple SIPP and test systems, different scenarios like uac, uas, msg...),
with a prep_offset of two seconds.
-Xavier
________________________________
From: Dushyant Dhalia [mailto:[email protected]]
Sent: vendredi 17 juillet 2009 7:33
To: Simonart, Xavier
Cc: [email protected]
Subject: Re: [Sipp-users] Problem running IMS_Bench - User_Pool getting
exhausting
Just wanted to know, what can be the minimum value of prep_offset and what
should be optimum value i.e. is there any relation between prep_offset and
number of users and call rate.
if prep_offset = f(x,y) then, what is x, y and f()?
Dushyant
Simonart, Xavier wrote:
Hi,
The problem, I think, is that the "calls" take longer than expected, when you
take everything into account.
The first call is initiated at 12:15:29.202 (in manager.log, O=2 means two
calls initiated). The first successful call is at 12:15:53.205 (S = 1).
So, that call took 24 (+/- 2) seconds (in fact, nothing guarantees that this is
the same call but we can say that the first call is either still busy or took
24 seconds). Those 24 seconds include the preparation time (prep_offset) and
the last 4 seconds (see pause milliseconds="4000" in uas).
Prep_offset is set to 20 seconds.
The error (pool empty) occurs at 12:15:32.193, at which time there are already
five calls initiated, and none completed. Five calls mean ten users (1 uac and
1 uas).
Hence, no more calls can be initiated (as there were ten users registered).
You should probably reduce the prep_offset (20 seconds looks very large to me -
I never saw it larger than 2 seconds). And increase the number of users (even
with a prep_offset of 2 seconds, the time during which users are in pool 3
would be around 3 seconds for the uac and 7 seconds for the uas)
-Xavier
________________________________
From: Dushyant Dhalia [mailto:[email protected]]
Sent: jeudi 16 juillet 2009 9:04
To: [email protected]<mailto:[email protected]>
Subject: [Sipp-users] Problem running IMS_Bench - User_Pool getting exhausting
Hi All,
I am using IMS_Bench, the scenario is -
<!-- Pre-Registration phase - Default mode is "single step" -->
<run cps="1" max_calls="10" distribution="constant" sync_mode="off"
use_scen_max_ihs="no" max_global_ihs="95" stats="1000">
<scenario name="ims_reg" ratio="100"/>
</run>
<run cps="0" duration="3"/> <!-- Sleep -->
<!-- Stir phase to warm up the SUT -->
<run cps="1" max_calls="3600" distribution="constant" use_scen_max_ihs="no"
max_global_ihs="95" stats="2000" report="no">
<scenario name="ims_uac" ratio="100"/>
</run>
<!-- Actual benchmark phase -->
<run cps="1" max_calls="10" distribution="constant" use_scen_max_ihs="no"
max_global_ihs="95" stats="2000">
<scenario name="ims_dereg" ratio="100"/>
</run>
Call holdup time is 1 second. Ideally before every call sipp should move a user
from user_pool[2] (registered users) to user_pool[3] (user in call) and after
successful completion of the call sipp should move this user back to the
user_pool[2] so that new calls can be initiated by "this user". But it seems
Sipp is not honouring this scenario (moving user from pool [3] to pool[2])
after successul completion of the scenario.
The error that i get is 2009-07-16 12:15:32.193: User pool id[2] is empty ->
Quitting!.
Can anybody suggest, what could be the error.
Thanks,
Dushyant P S Dhalia
---------------------------------------------------------------------
Intel Corporation NV/SA
Rond point Schuman 6, B-1040 Brussels
RPM (Bruxelles) 0415.497.718.
Citibank, Brussels, account 570/1031255/09
This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
---------------------------------------------------------------------
Intel Corporation NV/SA
Rond point Schuman 6, B-1040 Brussels
RPM (Bruxelles) 0415.497.718.
Citibank, Brussels, account 570/1031255/09
This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
------------------------------------------------------------------------------
Enter the BlackBerry Developer Challenge
This is your chance to win up to $100,000 in prizes! For a limited time,
vendors submitting new applications to BlackBerry App World(TM) will have
the opportunity to enter the BlackBerry Developer Challenge. See full prize
details at: http://p.sf.net/sfu/Challenge
_______________________________________________
Sipp-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sipp-users
