Hey Adrian,
Even though sindelka already touched on it, if you're having problems
*maintaining* concurrent calls it's most likely because you do not have a
sufficient *call holding time* set in your scenario.
To understand the relationship between the duration (or "flight time" as
sendelka put it) and the other load parameters check out the most basic
erlang formula:
https://en.wikipedia.org/wiki/Erlang_%28unit%29#Traffic_measurements_of_a_telephone_circuit
If you look at that algebraic relationship you can see that if your holding
time is not long enough then erlangs will of course also be low no matter
what call rate you choose.
Hope this helps!
Tyler Goodlet
On Mon, Mar 21, 2016 at 12:18 PM, sindelka <sinde...@ttc.cz> wrote:
> Hi Adi,
>
> in sipp, a "call" is a single pass through the from start to end,
> regardless whether it is an actual call-controlling (INVITE-200-BYE-200)
> dialog, a REGISTER-401-REGISTER-200 sequence, or e.g. a
> SUBSCRIBE-200-NOTIFY-200-... dialog.
>
> So in your register_client.xml, the "call" ends by receiving the 200
> response to the Register with authentication header.
>
> The number of balls in the air is determined by total flight time of a
> single ball (in your case, Σt = twice the network round-trip time plus
> Kamailio's processing time of non-authenticated and of authenticated
> REGISTER) and the number of balls you manage to throw until the first one
> falls back. If the Kamailio is quick enough and the network path is short
> enough, you may actually get just tens of really concurrent calls for 100
> calls in 80 ms. So I'd recommend to track the time of the first REGISTER
> and of the 200 response to the 2nd one (either using the sipp logging
> facility or using tcpdump/Wireshark) and modify -r and -rp depending on the
> result (like -r 1000 -rp Σt).
>
> Also think about the time it takes to send a message through a network
> interface of a given bitrate: a 1250-byte message represents about 10000
> bits to be sent, so on a 100 Mbit/s Ethernet, you cannot send more than 10
> such messages per millisecond, i.e. 800 in 80 milliseconds, if there is no
> other traffic on the interface.
>
> If you say that -r X -rp Y -l Z -m T doesn't do anything, try to add the
> parameters one-by-one, left-to-right, to find out which one breaks it, and
> then start omitting them one by one (-r -rp, -r -rp -l, -r -rp -l -m, -r -l
> -m, ...) to see if some combinations are not a problem, rather than
> individual parameters.
>
> This sentence is totally cryptic to me, it seems you were in a hurry:
>
> Also is this test to "create 100 calls during each 80 milliseconds up to
> a limit of 1000
> calls in progress" means that the calls, here register method calls not
> ths calls calls,
>
> are keep-alive simultaneous till the 2000 one ends?
>
>
> If what I wrote is not the answer, please translate the sentence.
>
> P.
>
>
> Dne 21.3.2016 v 16:00 adrian Albu napsal(a):
>
> Díky Sindelka,
>
> ./sipp -sf register_client.xml -inf user21.csv -i local_ip
> kamailio_ip:k_port -trace_err -r 100
>
> is kind of working using
>
> <?xml version="1.0" encoding="ISO-8859-1" ?>
> <scenario name="register_client">
> <send retrans="500">
> <![CDATA[
> REGISTER sip:[remote_ip] SIP/2.0
> Via: SIP/2.0/[transport] [local_ip]:[local_port];branch=[branch]
> From: <sip:[field0]@[field1]>;tag=[call_number]
> To: <sip:[field0]@[field1]>
> Call-ID: [call_id]
> CSeq: 1 REGISTER
> Contact: sip:[field0]@[local_ip]:[local_port]
> Max-Forwards: 5
> Expires: 1800
> User-Agent: SIPp/Linux
> Content-Length: 0
> ]]>
> </send>
> <recv response="401" auth="true">
> </recv>
> <send retrans="500">
> <![CDATA[
> REGISTER sip:[remote_ip] SIP/2.0
> Via: SIP/2.0/[transport] [local_ip]:[local_port];branch=[branch]
> From: <sip:[field0]@[field1]>;tag=[call_number]
> To: <sip:[field0]@[field1]>
> Call-ID: [call_id]
> CSeq: 2 REGISTER
> Contact: sip:[field0]@[local_ip]:[local_port]
> [field2]
> Max-Forwards: 5
> Expires: 1800
> User-Agent: SIPp/Linux
> Content-Length: 0
> ]]>
> </send>
> <recv response="200">
> </recv>
> </scenario>
>
> here is the result:
>
> ------------------------------ Scenario Screen -------- [1-9]: Change
> Screen --
> Call-rate(length) Port Total-time Total-calls Remote-host
> 100.0(0 ms)/1.000s 5060 20.05 s 2005
> kamailio_ip:k_port(UDP)
>
> 2 new calls during 0.021 s period 1 ms scheduler resolution
> 1 calls (limit 300) Peak was 1 calls, after 0 s
> 0 Running, 2008 Paused, 2 Woken up
> 0 dead call msg (discarded) 0 out-of-call msg (discarded)
> 3 open sockets
>
> Messages Retrans Timeout
> Unexpected-Msg
> REGISTER ----------> 2005 0 0
> 401 <---------- 2005 0 0 0
> REGISTER ----------> 2005 0 0
> 200 <---------- 2004 0 0 0
> ------------------------------ Test Terminated
> --------------------------------
>
>
> ----------------------------- Statistics Screen ------- [1-9]: Change
> Screen --
> Start Time | 2016-03-21 15:50:31.598871 1458571831.598871
> Last Reset Time | 2016-03-21 15:50:51.632419 1458571851.632419
> Current Time | 2016-03-21 15:50:51.654458 1458571851.654458
>
> -------------------------+---------------------------+--------------------------
> Counter Name | Periodic value | Cumulative value
>
> -------------------------+---------------------------+--------------------------
> Elapsed Time | 00:00:00:022000 | 00:00:20:055000
> Call Rate | 90.909 cps | 99.975 cps
>
> -------------------------+---------------------------+--------------------------
> Incoming call created | 0 | 0
> OutGoing call created | 2 | 2005
> Total Call created | | 2005
> Current Call | 1 |
>
> -------------------------+---------------------------+--------------------------
> * Successful call | 2 | 2004*
> Failed call | 0 | 0
>
> -------------------------+---------------------------+--------------------------
> Call Length | 00:00:00:003000 | 00:00:00:003000
> ------------------------------ Test Terminated
> --------------------------------
>
> But if I want to put more params to stress it a bit, like
>
> ./sipp -sf register_client.xml -inf user21.csv -i local_ip
> kamailio_ip:k_port -trace_err -r 100 -rp 80 -l 1000 -m 2000
>
> is doing nothing.
>
> Also is this test to "create 100 calls during each 80 milliseconds up to
> a limit of 1000
> calls in progress" means that the calls, here register method calls not
> ths calls calls,
>
> are keep-alive simultaneous till the 2000 one ends?
>
> I guess the ones from -r flag, here -r 100 , would be concurrent, but
> what is the default for the time till the sip session expires?
>
> I want to be sure I have an minimum number of sessions for a while to see
> it works. Maybe your explanation is clear but all these is new to me,
>
> I want to learn it but maybe am a bit impatient :)
>
> Thanks again,
> Adrian
> -----------------------------------------------------------------------------------------
>
> Take the power back!
>
>
> On Monday, 21 March 2016, 10:38, sindelka <sinde...@ttc.cz>
> <sinde...@ttc.cz> wrote:
>
>
> Bună ziua Adi,
>
> look at
> http://sipp.sourceforge.net/doc/reference.html#Online+help+%28-h%29
>
> with special care about the meaning of -r, -rp, -l and -m command line
> parameters.
>
> Example:
>
> sipp -sf your_scenario_name.xml -r 100 -rp 80 -l 1000 -m 2000
>
> will create 100 calls during each 80 milliseconds up to a limit of 1000
> calls in progress. Terminated calls will be replaced by new ones until the
> total number of calls will reach 2000. After that, no new calls will be
> added when the running ones finish. The scenario execution ends when the
> last call finishes.
>
> This is unrelated to the contents of the injection file and/or to whether
> it exists at all.
>
> The limitations of your hardware may cause the actual performance to be
> lower, especially if you are going to send RTP as part of the load test.
>
> P.
>
>
> Dne 21.3.2016 v 9:47 adrian Albu napsal(a):
>
> Hi,
>
> I was searching online for a way to test simultaneous sip sessions,
>
> for example 1000 connected on it at the same time.
>
> could not find an example,
>
> the only starting point I found was:
>
> <http://comments.gmane.org/gmane.comp.telephony.sipp.user/3934>
> http://comments.gmane.org/gmane.comp.telephony.sipp.user/3934
>
> Is there an example somewhere?
>
> Thanks,
> Adrian
>
>
> -----------------------------------------------------------------------------------------
> Take the power back!
>
>
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn
> more.http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
>
>
>
> _______________________________________________
> Sipp-users mailing
> listSipp-users@lists.sourceforge.nethttps://lists.sourceforge.net/lists/listinfo/sipp-users
>
>
>
>
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
>
> _______________________________________________
> Sipp-users mailing list
> Sipp-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/sipp-users
>
>
>
>
>
> ------------------------------------------------------------------------------
> Transform Data into Opportunity.
> Accelerate data analysis in your applications with
> Intel Data Analytics Acceleration Library.
> Click to learn more.
> http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
> _______________________________________________
> Sipp-users mailing list
> Sipp-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/sipp-users
>
>
------------------------------------------------------------------------------
Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
http://pubads.g.doubleclick.net/gampad/clk?id=278785351&iu=/4140
_______________________________________________
Sipp-users mailing list
Sipp-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sipp-users
