You are here:   Support > Wiki
Register   |  Login
 Wiki Minimize

VoIP tester


Voip test setttings

The default settings are ok for basic testing. You have to change only a few values after your needs.
Trace Level:
Meaningful values are from 0 (min loglevel) to 6 (max log level)
The logs are written to a logfile.

Display trace level:
How much details will be displayed on the application log section. Must be a value lower or equal with "trace level".

Codec preference:
For stress tests you should use PCMU or PCMA (otherwise the test pc's will have to big load)

Recording:
Sent and received media are stored in the "recorded" directory. You can decode them with the attached helper application (Converter.exe)
2 different files can be generated for one call. One for send and one for receive. You can mix this 2 file to a single stereo file with the converter application.

Converter.exe input parameters:
1. input file1
2. input file2 or "x" if no second file
3. output file name (can be empty)

Example 1:
Converter.exe "2222_2009_08_19_1195463507_de.pu" "2222_2009_08_19_1195463507_en.pu" "out.wav"

Example 2:
Converter.exe "2222_2009_08_19_1195463507_de.pu" x  out.wav

Example 3:
Converter.exe "2222_2009_08_19_1195463507_de.pu"

Audio settings:
If you uncheck the "Enable Audio" checkbox, then no rtp will be sent (but rtp will be still accepted)
The drop down list near the "enable audio" checkbox can have the following values:
-empty: in this case the audio cards will be used (not suitable for testing with multiple calls)
-Fake: random payload will be sent: the best option for stress testing
-file name: will playback the selected file

Send presence
Enable/disable presence (SUBSCIEBE/NOTIFY, PUBLISH)

On the network and authentication section the most important settings are the server ip and port.

If you would like to have username/password based authentication on your server, then you should set the username and password field correctly.
The username are also used as an A number for the outgoing calls.

QoS settings:
Enter any value between 1 and 63 or leave it empty (empty means critical priority + low delay for media and no qos for signaling)

Call generation settings:
All call count:
Total number of generated calls.

Max simultaneous calls:
max number of calls in the same time

New call start interval:
how fast new calls are generated (random between min and max)

Call duration:
Call duration will have a random value between min max.

Call prefix:
Outgoing called (B) number prefix or full number

Call num length:
the software will generate random digits after the "call prefix" digits.

For example if you need a fix called number as "11111" then you have to put "11111" in the call prefix and set the called number length to 5 (both min and max)

For call termination settings all values are generated randomly between the min-max values specified.
Ring time means after how much second the module will pick up the call.
Speech length: after how much second the module will disconnect the call
ASR: if not 100 then will randomly drop some calls to match the desired ASR value.

Average call duration and ASR are influenced by both the call generation and termination module.

Usually you will have to start the call termination module on a pc and the call generation module on another pc.
On the server you can identify this endpoints by address (ip:port) or after the sip username.
You should configure your server to route all calls to the "call termination module".
 
Factors that have big impact on performance are:
-max simoultan calls
-new call start interval
-codec (will affect only the clients -not affecting the server, so no reason to use g729 or g723 for tests)
-audio enabled/disabled 
-audio type: rec-playback device/fake/from file  (will affect only the clients -not affecting the server, so no reason to use other then "Fake")

To make an usual call (like from any other softphone):
+Check the "enable audio".
-The dropdown list near enable audio must be empty

To measure network rtp delay:
-set fake audio for the  call generation module
-disable audio on the call termination module

For stress testing the recommended values are:
-uncheck recording
-check enable audio end select "Fake"
-adjust the "max simoultan calls" and the "new call start interval" after your needs.
-during the test you can make a manual call (from a different softphone or ipphone) to check how is the call quality under high server load.

The most important statistics are displayed on the user interface.
For more details you can check the logfile. After every connected calls there are detailed statistics (if the loglevel is at least 3).
Example:
rtp statistics:
  session duration: 30 sec
  all packets: 341
  codec: G729
  average payload size: 20 bytes/packet (20 msec)
  average payload size (raw): 320 bytes/packet (20 msec)
  normal packets: 308 (99%)
  not processed packets: 0 (0%)
  lost packets from peer: 3 (1%)
  packet loss count: 1, average packet loss count: 32
  recovered packets: 0 (0%), resending: 0, resentout: 0
  out of order packets: 0 (0%), duplicated: 0 (0%)
  marked: 0 (0%), real: 0 (0%), silence: 0 (0%), dtmf: 0 (0%)
  ssrc change count: 0 (0%)
  used jitter size: 127 msec, min/start/max: 10/127/300
  soft buffer count: 39
  average driver quee length: 24 (491 msec)
  roundtrip delay: unknown
  prebuffering count: 0 (real: 0), packets: 6 (1%), emtyswjittercount: 0
  replay count: 0 (0%), from hardware: 0 (0%), max repl: 1
  dropp count: 32 (9%)
  pckt arrivals: 5:38,10:5,20:181,30:15,40:83,60:0,80:0,110:0,160:0,220:0,300:0,500:1,900:0

This demo version has 5  simoultan call limit.

VoIP Stresstest Presentation

VoIP tester Download



 |  View Topic History  |