How reliable is the data from ttcp?

Unanswered Question
Apr 3rd, 2008


I have used ttcp between two Cisco switches to test bandwidth capabilities on a 30 Mb circuit I have. The readings are what can only be described as disappointing. Should I assume that the ttcp numbers are good?

Thought I'd ask the experts.

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 4 (1 ratings)
Joseph W. Doherty Thu, 04/03/2008 - 16:29

Tftp transmission rate appears very sensitive to latency. Also, if you're actually reading/writing to flash, tftp transmission rate might reflect more of that rather than actual link capabilities. You should be able to obtain a more accurate bandwidth cabability measurement from a bandwidth test tool between two hosts.

cisco24x7 Thu, 04/03/2008 - 17:21

Iperf is the best tool for testing bandwidth

bar none, especially on Linux and Solaris


CCIE Security

lrian Sat, 04/05/2008 - 12:03

The ttcp numbers are good. But you didn't post the output from ttcp or the round trip time between the switches so dunno if you would have been able to use the full bandwidth or no. The max thruput you're going to get with tcp is the window size divided by the round trip time. Assuming an 8K window and ping reporting a round trip time of 5ms, the absolute best thruput you'd get is (8192*8) / .005 = about 13Mb/s

I don't remember the parameters for ttcp off the top of my head, but you need to set the receive window size to at least bandwidth * round trip time. Assuming a 5ms RTT between your switches and a 30Mb/s link, you'd need to have at least a (30,000,000 / 8) * .005 = 18,750 byte receive window to use the full bandwidth.

@EngZiad@ Thu, 01/27/2011 - 01:24

Dear All,

Can we use ttcp to measure 1 Gbps links between 2 Cisco L3  3750G switches?

now if i have 1 gbps link .Does this means i should use a receivewindow == 1,000,000,000 / 8 * RTT ( RTT for the 1 Gbps link is 1 msec)

so it gives me a rcvwndw=125,000 bytes?



This Discussion