UDP feedback mechanism?
I recently ran a set of UDP iperf tests between two directly connected 10GigEthernet servers, using
iperf 2.0.2. The connection in question was Ethernet encapsulated in SDH, and as part of the tests
we reduced the bottleneck capacity below the configured sending rate i.e. the source was instructed
to send at 2 Gbps, and then, during the test, the capacity of the SDH circuit was reduce to 1 Gbps.
What I expected to happen was that the receiver would start to see an amount of packet loss (50%)
- what actually happened was that the the transmitter reduced its sending rate. Is this epxected
behaviour on the part of iperf? I've seen a mail in the archive which refers to iperf's own
protocol running on top of UDP, and I guess this has somehow been invoked here, but is that by
design? If so is there some way I can force a high sending rate, regardless of any feedback mechanism?
Also, I know there is a patch for iperf to correct the UDP re-order count error - are there any
other patches I should be applying to correct known bugs? Perhaps there is a list somewhere of all
well-known patches?
kind regards
Toby
--
______________________________________________________________________
Toby Rodwell
Network Engineer
DANTE - www.dante.net
Tel: +44 (0)1223 371 300
Fax: +44 (0)1223 371 371
Email: toby.rodwell --at-- dante.org.uk
PGP Key ID: 0xB864694E
City House, 126-130 Hills Road
Cambridge CB2 1PQ
UK
_____________________________________________________________________