Re: UDP - Iperf Loss ??


Hello Marc.

thank your for your prompt reply.

let me answer in between...

Marc Herbert wrote:

> On Thu, 29 Jul 2004, Marcelo Maraboli wrote:
> 
> 
>>Iperf:
>>PC1: iperf -c IP-PC2 -u -w 200k -l 12 -n 1785708 -b 100m
>>	(which would lead to 148809 frames/s at 100mbps)
>>PC2: iperf -s -u -w 200k
>>
>>I know that 100m at UDP is different than 100m at Ethernet,
>>but I also tested at 90m.
> 
> 
> Headers in this case are 42 bytes long (just tested on loopback). So
> with -l 12, packets are 54 bytes long, so you can get at most...
> 22Mb/s. Quite different from 100M (or event 90M) indeed.

I do not agree (but maybe I´m wrong ;):

Ethernet header: 6+6+2 = 14 bytes
IP header: 20 bytes
udp header: 8 bytes
UDP data-payload: 12 bytes
Ethernet padding: 6 bytes
CRC FCS: 4 bytes
----
TOTAL Ethernet packet length: 64 bytes

,which is the minimum Ethernet packet length.

Your 42 header bytes only include: 14 + 20 + 8 = 42

>>STRANGE:
>>I check the frame counter on the switch, and the receiving
>>port does receive 148792 frames from PC1. The switch
>>outputs (on PC2 port) the amount of 148793 frames. (which
>>is saying that the switch is non-blocking or did not lose
>>any frames), but Iperf reports the following in PC2:
>>
>>time	   transfer    bw	jitter	  loss/total
>>0.0-2.3s   474 Kbytes  1.71Mbps  0.047ms  108341/148809 (73%)
> 
> 
> This is confusing: is the "148809" a rate (frames/s) or just a
> volume (frames) ?

It is not a rate, only total amount of packets that SHOULD be
received...

> 
> 
> 
>>- This tells me that PC2 did not receive the 148809 frames ??
> 
> 
> Right, it only received 73% of them.
> 

right!

> 
> 
>>- then how did it know there were 148809 frames sent to him ??
> 
> 
> iperf has its own (and convenient) protocol on top of UDP. Sequence
> numbers, ACKnowledgements, whatever...
> 

that explains the few extra packets... ;)

> 
> 
>>- WHY did Iperf take 2.3s to send 148809 packets if I told
>>it to do it at 100mbps ?? (Dlink could not handle it?)
> 
> 
> Maybe because you can get only 22Mb/s because of protocols
> headers/overheads ? And there also maybe some queueing here and
> there...
> 

Queueing in PC1-output, switch-input, switch-output and PC2-input.

The strange thing is that LOSS is only registered at PC2, because
Switch counters do not show any loss (error counters, drop counters).

Is this a Iperf flaw or a Operating System not tunned. ?

> Could you give more details about your computations and expectations ?
> (sorry, I am too lazy to re-do them by myself :-)
> 

I am using Iperf to test switches at the maximum Ethernet
spec at various packet lenghts, as stated in RFC 2544.

My problem/goal is:

how can I use Iperf to generate 148809 Ethernet packets/s
of a 64 bytes length at 100Mbps ??? (to test the switch)

also meaning: it should take 1 second (if the NIC allows it),
and 1 second to receive it (if the switch and Pc2-NIC allows it).

thank you,
-- 
Marcelo Maraboli Rosselott
Jefe Area de Redes           (Network & UNIX Systems Administrator)
Ingeniero Civil Electronico                   (Electronic Engineer)

Direccion Central de Servicios Computacionales (DCSC)
Universidad Tecnica Federico Santa Maria, Chile.
phone: +56 32 654237
mailto:marcelo.maraboli --at-- dcsc.utfsm.cl	http://elqui.dcsc.utfsm.cl/



Other Mailing lists | Author Index | Date Index | Subject Index | Thread Index