Lennert:
> Did you just look at 'top' for determining this? Time spent in
> interrupt context isn't counted in that figure :-)
I used 'vmstat'.
******* I am sorry to send too long mail.
Here, are some results. I used 'iperf' and 'nttcp' both. Before, you read
further in details, I got performance of around 520-580 Mbits/sec both in
UDP/TCP (using both iperf and ttcp). And one of my end machines is running
FreeBSD 4.5
Please read /Questions/ at the end of the results.
My observation during testing was, Linux box could not push more than 580
Mbits/sec ( I am using Interl 82543 Gigcards, P-III 1.26 GHz, 133 MHz bus
speed and 1 GB memory on all machines).
Setup: (/nttcp/)
nttcp sender : test-machine (OS: Linux-2.4.18)
bridge : br-machine (OS: Linux-2.4.18)
nttcp receiver : test-machine2 (OS: FreeBSD 4.5) // Please Note
l=local, 1=remote(receiver),
For man page of 'nttcp'
goto: http://www.leo.org/~elmar/nttcp/nttcp.1.html
1) UDP:
-n 252144 (that many number of pkt has been sent, each has 4096Bytes
long)
Bytes Real s CPU s Real-MBit/s CPU-Mbit/s Calls Real-C/s CPU-C/s
l 1032781824 14.37 4.30 575.1554 1921.4546 252147 17552.56 58638.8
1 945160192 14.37 1.51 526.3189 5012.7961 230753 16062.05 152979.1
br-machine cpu consumption went 14% (vmstat)
2) TCP:
-n 252144
Bytes Real s CPU s Real-MBit/s CPU-Mbit/s Calls Real-C/s CPU-C/s
l 1032781824 15.15 2.69 545.2294 3071.4701 252144 16639.08 93733.8
1 1032781824 15.15 2.36 545.1938 3505.5973 299033 19732.02 126876.9
br-machine cpu consumtion went 43% (vmstat)
Setup: (/iperf/)
iperf Server : test-machine2 (OS: FreeBSD 4.5)
bridge : br-machine (OS: Linux 2.4.18)
iperf Client : test-machine (OS: Linux 2.4.18)
1) UDP:
Server: iperf -s -u
Client: iperf -c 10.4.2.5 -u -b 900m -t 10
Server Result:
Interval: 0.0 - 10.0 sec
Transfer: 691 MBytes
B/W : 580 Mbits/sec
Jitter : 0.049 ms
Lost/Total : 0/454456 (0.0%)
Client Result:
Interval: 0.0 - 10 sec
Transfer: 691 MBytes
B/w : 553 Mbits/sec
Sent : 493105 datagrams
bridge used 15% of cpu
2) TCP:
Server: iperf -s (Default window 32KB)
Client: iperf -c 10.4.2.5 (Default window 16KB) -t 40
Client/Server Result:
Interval: 0.0 - 40 sec
Transfer: 2.6 GBytes
B/W : 560 Mbits/sec (Client 534 Mbits/sec)
bridge used 44% of cpu
Questions:
1) Why br-machine (running bridge) cpu load went high during TCP (upto
44%) and remained low(upto 18%) during UDP ?
2) Why Linux machine could not push more than 580 Mbits/sec. (even i gave
iperf -c -u -b 900 m). While FreeBSD could push 950 Mbits/sec. (BEWARE:
Here I am not saying that Linux is bad or whatever...). Is it a driver
issue ?
3) Currently i am only doing bridging, but i am planning to do more on top
of bridging (e.g., shaping, policing based on different attributes). So,
pkt might goto Layer 3 in that case. Will performance degrade in that case
?
4) Lennert:
> Yeah, odd. I get full wire speed (120 megabytes/sec) over TCP
> between a dual Athlon 1.2ghz and a Pentium III 1.2ghz (where
> the athlon is the transmitting host, network card interrupt
> bound to CPU#0).
If it is possible, could you tell me what else did you use ?
(Specification of machine/card etc., application which measured
performance etc.)
Many Thanks,
-Kunal
_______________________________________________
Bridge mailing list
[EMAIL PROTECTED]
http://www.math.leidenuniv.nl/mailman/listinfo/bridge
