Skip to Content.
Sympa Menu

ndt-users - Re: gigabit slower than fast-e

Subject: ndt-users list created

List archive

Re: gigabit slower than fast-e


Chronological Thread 
  • From: Peter Van Epp <>
  • To:
  • Subject: Re: gigabit slower than fast-e
  • Date: Thu, 6 Mar 2008 08:09:58 -0800

On Thu, Mar 06, 2008 at 12:04:18AM -0500,

wrote:
> I'm troubleshooting data transfers between New Jersey and California over
> internet2. The slowest link is 155 Mbps, rtt is 90 ms, servers and
> switches are gigabit.
>
> Tuning tcp buffers only gets the throughput to 11.9 Mbits/sec. If I change
> the switch port to 100 Full, the throughput increases to 84.4 Mbits/sec.
>
> ndt reports packet loss and duplicate acks when the connection is gigabit,
> the 100 full connection has no packet loss.
>
> The relevant ndt output is below. iperf also backs up these results.
>
> Is there a known problem with gig switch negotiation? No tcp buffer
> setting improves the performance when gigabit.
>
> Thanks
>
> ndt:
>
<snip>

Are your window sizes boosted? You may be seeing tcp window exhaustion
somewhere on the link if you are using the default 65K windows (or 8K in some
impelementations) they may not be big enough on the gig link (but are at 100).

in Statistics:

...
RFC 1323 Window scaling: ON

in More details:

...
WinScaleRcvd: 3
WinScaleSent: 8

If thats all OK then the answer probably is buffers. Our server (the
one with window scaling at 8) can achieve Gig across about 3000 miles here
in Canada (which likely doesn't help you though :-)) with jumbo frames to
a similarly tuned system on the other end. We have seen as low as 35 megabits
to an untuned system on the other end on the same path.

Peter Van Epp / Operations and Technical Support
Simon Fraser University, Burnaby, B.C. Canada



Archive powered by MHonArc 2.6.16.

Top of Page