DHCP transactions per second

Ian Anderson ian.anderson at clearwire.com
Mon Jun 18 16:45:17 UTC 2007


Glenn Satchell wrote:
>> Date: Mon, 18 Jun 2007 09:35:56 +0100
>> To: dhcp-users at isc.org
>> From: Simon Hobson <dhcp1 at thehobsons.co.uk>
>> Subject: Re: DHCP transactions per second
>>
>> Ian Anderson wrote:
>>     
>>> With further investigation it appears that we are getting multiple
>>> requests for each client.  We have several ip helper  addresses (foundry
>>> switchs) in our network, and I believe we may have a misconfiguration
>>> somewhere.
>>>
>>> For example, if a client has never powered up on our network before it
>>> performs the typical DISCOVER/OFFER/REQUEST/ACK  Because all of those
>>> are broadcast the dhcp server is receiving the client messages 4 times. 
>>> (and responding 4 times as well) 4 DISCOVERS / 4 REQUESTS.  I know dhcp
>>> can cope with this, and the server will ACK the first REQUEST it
>>> receives from any given client.   Even though dhcp can handle this, is
>>> this a significant source of additional load on the server?
>>>       
>> More or less, each request adds one unit of extra work.
>>
>>     
>>> Are duplicate broadcasts treated any differently by dhcp than receiving just
>>> one broadcast?
>>>       
>> No
>>
>>     
>>> Or in other words, does dhcp know the difference between a normal 
>>> broadcast and one that is a duplicate?
>>>       
>> How can it ? They are simply relayed packets and will be identical 
>> other than the GIAddr - there is nothing to say "this is a duplicate".
>>
>>     
If this is the case, and DHCP responds to every request it receives, 
what happens to the DHCPOFFER that are never ACK'd by the client. If 
DHCP transmits 4 DHCPOFFERS and a client ACK's one of them, what happens 
to the other 3?
>>> I would imagine in order to get an accurate representation of your dhcp
>>> servers transactions per second you would need to fix whatever is
>>> causing the duplicate messages.
>>>       
>
> In terms of redundancy having more than one relay agent could be
> considered "good practice". Perhaps reconfiguring your relay agents so
> that between your multiple foundry devices you have only two for each
> subnet.
>
> regards,
> -glenn
>
>   



More information about the dhcp-users mailing list