Generate Udp Traffic Windows
UDP - Sender/Reciever. Send/Receive allows you to send UDP datagrams and if selected, to listen for a response bound to the port on local host. Receive only listens indefinitely for UDP packets and outputs the result in the log section. Send only allows you to send UDP datagrams to specified host on given port - Outputs the sender (destination).
- Packet Sender is a free utility to for sending / receiving of network packets. Support for TCP, UDP, and SSL. Ad: Pay Down Calc will help you calculate your loan.
- WAN Killer software is built to be a random traffic generator that sends network traffic to the IP address and desired hostname. Once you install Engineer’s Toolset, you can find WAN Killer under the Diagnostics folder.
Windows Test Udp Port
Hi guys,
I have the following scenario: 2 virtual machines connected through 3 routers in dynamips. My VMs have WinXP installed on them.
My goal is to generate as much traffic as i can by means of iperf, so I can measure the TCP throughput. How can I do that? I know one way is to modify TCP windows size(socket buffer size).
If I use the default setting (default TCP windows size is 8KB), I have the following output:
C:>iperf.exe -c 192.168.40.128 -i 1
------------------------------------------------------------
Client connecting to 192.168.40.128, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[1916] local 192.168.30.128 port 1250 connected with 192.168.40.128 port 5001
[ ID] Interval Transfer Bandwidth
[1916] 0.0- 1.0 sec 176 KBytes 1.44 Mbits/sec
[1916] 1.0- 2.0 sec 224 KBytes 1.84 Mbits/sec
[1916] 2.0- 3.0 sec 200 KBytes 1.64 Mbits/sec
[1916] 3.0- 4.0 sec 208 KBytes 1.70 Mbits/sec
[1916] 4.0- 5.0 sec 144 KBytes 1.18 Mbits/sec
[1916] 5.0- 6.0 sec 128 KBytes 1.05 Mbits/sec
[1916] 6.0- 7.0 sec 136 KBytes 1.11 Mbits/sec
[1916] 7.0- 8.0 sec 208 KBytes 1.70 Mbits/sec
[1916] 8.0- 9.0 sec 192 KBytes 1.57 Mbits/sec
[1916] 9.0-10.0 sec 144 KBytes 1.18 Mbits/sec
[1916] 0.0-10.1 sec 1.73 MBytes 1.43 Mbits/sec
C:>iperf.exe -s -i 1 -t 10
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[1884] local 192.168.40.128 port 5001 connected with 192.168.30.128 port 1250
[ ID] Interval Transfer Bandwidth
[1884] 0.0- 1.0 sec 184 KBytes 1.51 Mbits/sec
[1884] 1.0- 2.0 sec 216 KBytes 1.77 Mbits/sec
[1884] 2.0- 3.0 sec 200 KBytes 1.64 Mbits/sec
[1884] 3.0- 4.0 sec 208 KBytes 1.70 Mbits/sec
[1884] 4.0- 5.0 sec 152 KBytes 1.25 Mbits/sec
[1884] 5.0- 6.0 sec 120 KBytes 983 Kbits/sec
[1884] 6.0- 7.0 sec 136 KBytes 1.11 Mbits/sec
[1884] 7.0- 8.0 sec 208 KBytes 1.70 Mbits/sec
[1884] 8.0- 9.0 sec 192 KBytes 1.57 Mbits/sec
[1884] 0.0-10.0 sec 1.73 MBytes 1.45 Mbits/sec
which doesn't help me.
Now if I change TCP windows size by following the rule: bandwidth*RTT/8, I get: 100Mbit/s*0.1s/8= 1.25MBytes. Because RTT varies, I chose -w 1.5M.
C:>iperf.exe -c 192.168.40.128 -i 1 -w 1.5M
------------------------------------------------------------
Client connecting to 192.168.40.128, TCP port 5001
TCP window size: 1.50 MByte
------------------------------------------------------------
[1916] local 192.168.30.128 port 1254 connected with 192.168.40.128 port 5001
[ ID] Interval Transfer Bandwidth
[1916] 0.0- 1.0 sec 1.53 MBytes 12.8 Mbits/sec
[1916] 1.0- 2.0 sec 32.0 KBytes 262 Kbits/sec
[1916] 2.0- 3.0 sec 112 KBytes 918 Kbits/sec
[1916] 3.0- 4.0 sec 192 KBytes 1.57 Mbits/sec
[1916] 4.0- 5.0 sec 208 KBytes 1.70 Mbits/sec
[1916] 5.0- 6.0 sec 176 KBytes 1.44 Mbits/sec
[1916] 6.0- 7.0 sec 144 KBytes 1.18 Mbits/sec
[1916] 7.0- 8.0 sec 168 KBytes 1.38 Mbits/sec
[1916] 8.0- 9.0 sec 40.0 KBytes 328 Kbits/sec
[1916] 9.0-10.0 sec 128 KBytes 1.05 Mbits/sec
[1916] 0.0-21.3 sec 2.71 MBytes 1.07 Mbits/sec
C:>iperf.exe -s -i 1 -t 10 -w 1.5M
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 1.50 MByte
------------------------------------------------------------
[1884] local 192.168.40.128 port 5001 connected with 192.168.30.128 port 1253
[ ID] Interval Transfer Bandwidth
[1884] 0.0- 1.0 sec 217 KBytes 1.78 Mbits/sec
[1884] 1.0- 2.0 sec 288 KBytes 2.36 Mbits/sec
[1884] 2.0- 3.0 sec 272 KBytes 2.23 Mbits/sec
[1884] 3.0- 4.0 sec 264 KBytes 2.16 Mbits/sec
[1884] 4.0- 5.0 sec 175 KBytes 1.44 Mbits/sec
[1884] 5.0- 6.0 sec 192 KBytes 1.58 Mbits/sec
[1884] 6.0- 7.0 sec 304 KBytes 2.49 Mbits/sec
[1884] 7.0- 8.0 sec 272 KBytes 2.23 Mbits/sec
[1884] 8.0- 9.0 sec 248 KBytes 2.03 Mbits/sec
[1884] 9.0-10.0 sec 279 KBytes 2.29 Mbits/sec
[1884] 10.0-11.0 sec 305 KBytes 2.50 Mbits/sec
[1884] 11.0-12.0 sec 207 KBytes 1.69 Mbits/sec
[1884] 12.0-13.0 sec 168 KBytes 1.38 Mbits/sec
[1884] 13.0-14.0 sec 184 KBytes 1.51 Mbits/sec
[1884] 14.0-15.0 sec 272 KBytes 2.23 Mbits/sec
[1884] 15.0-16.0 sec 336 KBytes 2.76 Mbits/sec
[1884] 0.0-16.2 sec 3.96 MBytes 2.05 Mbits/sec
As you can see, in the first second on the client side, the throughput was 12.8 Mbits/s, but then it dramatically decreased. What could be the cause of bandwidth variations between client side and server side. RTT?
Then, there is a delay on the server side. It counted for 16 seconds, even if I specified -t 10. Why is that?
Do you know other methods that could help me to measure the throughput? What about -l option?
I also read about the default TCP send and receive buffer sizes of 8kBytes on Windows systems. These parameters can be tuned to maximize TCP/IP performance. What do you know about this? Do you think it is necessary in my case?
Thank you!
I have the following scenario: 2 virtual machines connected through 3 routers in dynamips. My VMs have WinXP installed on them.
My goal is to generate as much traffic as i can by means of iperf, so I can measure the TCP throughput. How can I do that? I know one way is to modify TCP windows size(socket buffer size).
If I use the default setting (default TCP windows size is 8KB), I have the following output:
C:>iperf.exe -c 192.168.40.128 -i 1
------------------------------------------------------------
Client connecting to 192.168.40.128, TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[1916] local 192.168.30.128 port 1250 connected with 192.168.40.128 port 5001
[ ID] Interval Transfer Bandwidth
[1916] 0.0- 1.0 sec 176 KBytes 1.44 Mbits/sec
[1916] 1.0- 2.0 sec 224 KBytes 1.84 Mbits/sec
[1916] 2.0- 3.0 sec 200 KBytes 1.64 Mbits/sec
[1916] 3.0- 4.0 sec 208 KBytes 1.70 Mbits/sec
[1916] 4.0- 5.0 sec 144 KBytes 1.18 Mbits/sec
[1916] 5.0- 6.0 sec 128 KBytes 1.05 Mbits/sec
[1916] 6.0- 7.0 sec 136 KBytes 1.11 Mbits/sec
[1916] 7.0- 8.0 sec 208 KBytes 1.70 Mbits/sec
[1916] 8.0- 9.0 sec 192 KBytes 1.57 Mbits/sec
[1916] 9.0-10.0 sec 144 KBytes 1.18 Mbits/sec
[1916] 0.0-10.1 sec 1.73 MBytes 1.43 Mbits/sec
C:>iperf.exe -s -i 1 -t 10
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 8.00 KByte (default)
------------------------------------------------------------
[1884] local 192.168.40.128 port 5001 connected with 192.168.30.128 port 1250
[ ID] Interval Transfer Bandwidth
[1884] 0.0- 1.0 sec 184 KBytes 1.51 Mbits/sec
[1884] 1.0- 2.0 sec 216 KBytes 1.77 Mbits/sec
[1884] 2.0- 3.0 sec 200 KBytes 1.64 Mbits/sec
[1884] 3.0- 4.0 sec 208 KBytes 1.70 Mbits/sec
[1884] 4.0- 5.0 sec 152 KBytes 1.25 Mbits/sec
[1884] 5.0- 6.0 sec 120 KBytes 983 Kbits/sec
[1884] 6.0- 7.0 sec 136 KBytes 1.11 Mbits/sec
[1884] 7.0- 8.0 sec 208 KBytes 1.70 Mbits/sec
[1884] 8.0- 9.0 sec 192 KBytes 1.57 Mbits/sec
[1884] 0.0-10.0 sec 1.73 MBytes 1.45 Mbits/sec
which doesn't help me.
Now if I change TCP windows size by following the rule: bandwidth*RTT/8, I get: 100Mbit/s*0.1s/8= 1.25MBytes. Because RTT varies, I chose -w 1.5M.
C:>iperf.exe -c 192.168.40.128 -i 1 -w 1.5M
------------------------------------------------------------
Client connecting to 192.168.40.128, TCP port 5001
TCP window size: 1.50 MByte
------------------------------------------------------------
[1916] local 192.168.30.128 port 1254 connected with 192.168.40.128 port 5001
[ ID] Interval Transfer Bandwidth
[1916] 0.0- 1.0 sec 1.53 MBytes 12.8 Mbits/sec
[1916] 1.0- 2.0 sec 32.0 KBytes 262 Kbits/sec
[1916] 2.0- 3.0 sec 112 KBytes 918 Kbits/sec
[1916] 3.0- 4.0 sec 192 KBytes 1.57 Mbits/sec
[1916] 4.0- 5.0 sec 208 KBytes 1.70 Mbits/sec
[1916] 5.0- 6.0 sec 176 KBytes 1.44 Mbits/sec
[1916] 6.0- 7.0 sec 144 KBytes 1.18 Mbits/sec
[1916] 7.0- 8.0 sec 168 KBytes 1.38 Mbits/sec
[1916] 8.0- 9.0 sec 40.0 KBytes 328 Kbits/sec
[1916] 9.0-10.0 sec 128 KBytes 1.05 Mbits/sec
[1916] 0.0-21.3 sec 2.71 MBytes 1.07 Mbits/sec
C:>iperf.exe -s -i 1 -t 10 -w 1.5M
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 1.50 MByte
------------------------------------------------------------
[1884] local 192.168.40.128 port 5001 connected with 192.168.30.128 port 1253
[ ID] Interval Transfer Bandwidth
[1884] 0.0- 1.0 sec 217 KBytes 1.78 Mbits/sec
[1884] 1.0- 2.0 sec 288 KBytes 2.36 Mbits/sec
[1884] 2.0- 3.0 sec 272 KBytes 2.23 Mbits/sec
[1884] 3.0- 4.0 sec 264 KBytes 2.16 Mbits/sec
[1884] 4.0- 5.0 sec 175 KBytes 1.44 Mbits/sec
[1884] 5.0- 6.0 sec 192 KBytes 1.58 Mbits/sec
[1884] 6.0- 7.0 sec 304 KBytes 2.49 Mbits/sec
[1884] 7.0- 8.0 sec 272 KBytes 2.23 Mbits/sec
[1884] 8.0- 9.0 sec 248 KBytes 2.03 Mbits/sec
[1884] 9.0-10.0 sec 279 KBytes 2.29 Mbits/sec
[1884] 10.0-11.0 sec 305 KBytes 2.50 Mbits/sec
[1884] 11.0-12.0 sec 207 KBytes 1.69 Mbits/sec
[1884] 12.0-13.0 sec 168 KBytes 1.38 Mbits/sec
[1884] 13.0-14.0 sec 184 KBytes 1.51 Mbits/sec
[1884] 14.0-15.0 sec 272 KBytes 2.23 Mbits/sec
[1884] 15.0-16.0 sec 336 KBytes 2.76 Mbits/sec
[1884] 0.0-16.2 sec 3.96 MBytes 2.05 Mbits/sec
As you can see, in the first second on the client side, the throughput was 12.8 Mbits/s, but then it dramatically decreased. What could be the cause of bandwidth variations between client side and server side. RTT?
Then, there is a delay on the server side. It counted for 16 seconds, even if I specified -t 10. Why is that?
Do you know other methods that could help me to measure the throughput? What about -l option?
I also read about the default TCP send and receive buffer sizes of 8kBytes on Windows systems. These parameters can be tuned to maximize TCP/IP performance. What do you know about this? Do you think it is necessary in my case?
Thank you!