[OpenSIPS-Users] Some results => Re: register performance with sipp

[Uwe Kastens]

Hi,

In my testing setup I am not able to reach more than 500 cps without
massive retransmits on db_mode=3. On db_mode=2 i reach with the same
setup 900cps. I tried with one sipp client and with two - it makes no
difference.

All hosts are on xen except the db-server.

My setup:

- 1 or 2 sipp instances (it makes no difference)
- 1 opensips 1.5.1 1024MB RAM
- freeradius with 8192 max_connections, and 25 sql socks against mysql
- opensips will register via radius at mysql
- usrloc db_mode=3 (only db) or db_mode=2


I generate user and password via pwgen and but them in radcheck db.

I use a REGISTER scenario with sipp and start it like:

 sipp -sf register.xml -inf user.csv xxx.xxx.de -m 5000 -r 500 -nd
                                 Messages  Retrans   Timeout
Unexpected-Msg
    REGISTER ---------->         5000      0         0
         401 <----------         5000      0                   0
    REGISTER ---------->         5000      36        0
         200 <----------         5000      0                   7
------------------------------ Test Terminated
--------------------------------


----------------------------- Statistics Screen ------- [1-9]: Change
Screen --
  Start Time             | 2009-06-22 09:21:17

  Last Reset Time        | 2009-06-22 09:21:27

  Current Time           | 2009-06-22 09:21:27

-------------------------+---------------------------+--------------------------
  Counter Name           | Periodic value            | Cumulative value
-------------------------+---------------------------+--------------------------
  Elapsed Time           | 00:00:00:202              | 00:00:10:218

  Call Rate              |    0.000 cps              |  489.333 cps

-------------------------+---------------------------+--------------------------
  Incoming call created  |        0                  |        0

  OutGoing call created  |        0                  |     5000

  Total Call created     |                           |     5000

  Current Call           |        0                  |

-------------------------+---------------------------+--------------------------
  Successful call        |        4                  |     5000

  Failed call            |        0                  |        0

-------------------------+---------------------------+--------------------------
  Call Length            | 00:00:00:137              | 00:00:00:015

------------------------------ Test Terminated
--------------------------------

2009-06-22 09:21:27: Discarding message which can't be mapped to a known
SIPp call:
SIP/2.0 401 Unauthorized
Via: SIP/2.0/UDP 10.20.139.62:5060;branch=z9hG4bK-18810-4611-2
From: <sip:aizeejenupigeeghefoo at xxx.xxx.de>;tag=4611
To:
<sip:aizeejenupigeeghefoo at xxx.xxx.de>;tag=a0ed7041c178b85c9908fb3ff756ca71.2356
Call-ID: 4611-18810 at 10.20.139.62
CSeq: 2 REGISTER
WWW-Authenticate: Digest realm


Uwe Kastens schrieb:
> Hi,
> 
> Thanks for the pointer. I should mabye learn to search on the mailing
> list :)
> 
> Do you remember what your hardware setup looked like? Esp I would be
> very interested in the hardware you used for the sipp requests.
> 
> Did you see any errors like this on your tests "Discarding message which
> can't be mapped to a known SIPp call"? I was wondering if this is caused
> by the opensips server or by some pkg sipp is loosing.
> 
> I think I would like to document some of the tests in the opensips wiki.
> 
> BR
> 
> Uwe
> 
> 
> Iñaki Baz Castillo schrieb:
>> El Domingo, 21 de Junio de 2009, Uwe Kastens escribió:
>>> Hello,
>>>
>>> Anybody experience with measuring REGISTER performance with sipp? I made
>>> some tests and I was wondering how many requests should be possible with
>>> opensips/sipps (radius against mysql).
>>>
>>> It looks like, that I can handle easily 500 REGISTER requests per sec on
>>> a XEN Domain (one for sipp and one for opensips), database is on
>>> dedicated quad-core server.
>>>
>>> I found out that my freeradius config caused some trouble (max_requests).
>>>
>>> What could I expect with that setup?
>> Hi, I did some tests about it (with no radius however):
>>
>>   http://lists.opensips.org/pipermail/users/2008-December/002074.html
>>
>>
> 
> 


-- 

kiste lat: 54.322684, lon: 10.13586