DHCP transactions per second

[Ian Anderson]

dhcperf works really well.  It allows you to test DHCPDISCOVER tps as 
well as RENEW tps.  It starts with 30 clients/sec and then keep working 
its way up.  The one downfall is that it is extremely picky about when 
it needs to receive a response from the server.  I believe it needs to 
receive a response within .1 seconds, if it doesn't it considers that a 
failure.  This is not a real world test by any means, as clients will 
retransmit their requests.  dhcperf does not take this into account. 



Ragnar Lonn wrote:
> Ian Anderson skrev:
>   
>> Math was never my strongest subject and I am hoping someone can offer 
>> advise as to wether or not this is the proper way to compute DHCP tps.  
>> I will be using a combination of the Ixia IXLoad test suite and 
>> Nominum's dhcperf tool to perform stress tests.   All advise is welcome.
>>   
>>     
>
>
> What is your experience of dhcperf?  I've had requests to include stress 
> testing functionality
> in my dhcptool test tool 
> (http://gatorhole.com/index.php?product=dhcp&lang=gb) but if
> there already are good open-source stress testing tools maybe I 
> shouldn't spend time on
> reinventing the wheel?
>
> Regards,
>
>   /Ragnar
>
>