Re: Client and server on same computer
On 06/01/2007 01:30 AM, Ben Greear wrote:
> Dag Bakke wrote:
>> On 05/31/2007 09:55 AM, Daniel Janzon wrote:
>
>>> This would be a dedicated machine. If the speed of the ethernet
>>> card(s) is significantly higher than the equipment under test,
>>> I don't see how it could be a problem using a single machine.
>>
>>
>> You may be in for a surprise. Any other traffic running over the same
>> PCI-bus as your network card or consuming CPU cycles will likely affect
>> your results negatively. What is the bandwidth of the bus the network
>> interface sits on, and what devices share that bus?
>>
>> Even modern hardware can have problems filling a 100mbps link with small
>> packets. (Too many interrupts to handle.) Certain GbE interfaces
>> struggles really hard to receive 300 Mbps for any packetsize, and will
>> definitely not be able to transmit that much.
>
> An Intel pro/1000 NIC on a PCI-X or PCI-E bus will have no problem
> sending or receiving 900+Mbps. Using pktgen or similar on Linux you
> can generate around 1 million (64k) packets per second using similar
> hardware.
>
> This is more than adequate to fully utilize a large amount of wifi
> equipment!
Yup. Intel gear is very good. Particularly on a fat bus. But there are
plenty non-Intel GbE-interfaces out there sitting on a 32bit PCI-bus.
And sadly, AFAICT only Intel motherboards got embedded Intel GbE
interfaces, (And some laptops.)
> To make sure it's not your test system that is a bottle-neck, just use
> loop-back ethernet cables instead of the network under test...
Exactly. Then run iperf with a short reporting interval and verify that
your throughput numbers are stable.
Dag B.