broken re(4)

[Gerrit Kühn]

On Thu, 12 Jun 2008 16:01:26 +0900 Pyun YongHyeon <pyunyh at gmail.com> wrote
about Re: broken re(4):

PY>  > I already did simple benchmarking by using "dd if=/dev/zero
PY>  > of=file" which gave me several 10s of MByte/s under all
PY>  > circumstances. Can you recommend one of the benchmarking programs
PY>  > for more detailed testing?

PY> Try netperf or iperf in ports/benchmark.

The machine in question as client:

------------------------------------------------------------
Client connecting to 130.75.117.1, TCP port 5001
TCP window size: 32.5 KByte (default)
------------------------------------------------------------
[  3] local 130.75.117.112 port 56513 connected with 130.75.117.1 port 5001
[  3]  0.0-10.0 sec    211 MBytes    177 Mbits/sec

On the server side:
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[  4] local 130.75.117.1 port 5001 connected with 130.75.117.112 port 56513
[  4]  0.0-10.3 sec    211 MBytes    172 Mbits/sec



As server:
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[  4] local 130.75.117.112 port 5001 connected with 130.75.117.1 port 53843
[  4]  0.0-10.1 sec    399 MBytes    331 Mbits/sec

On the client side:
------------------------------------------------------------
Client connecting to 130.75.117.112, TCP port 5001
TCP window size: 32.5 KByte (default)
------------------------------------------------------------
[  3] local 130.75.117.1 port 53843 connected with 130.75.117.112 port 5001
[  3]  0.0-10.1 sec    399 MBytes    331 Mbits/sec



Hm, being a server seems to work better? The machine on the other side is
also having a re-interface and usually transfers data with 20-30MByte/s.
The ITX machine I'm testing is using both of it's re-interfaces in a
lagg-configuration right now (laggproto loadbalance).

Is this the expected performance?


cu
  Gerrit