Gigabit speed test anybody?

Resent from my subscribed address. Hopefully this isn't a dupe to anybody.

You can try:


Rick Ernst wrote:


Thanks for the link. I don't currently have a Linux box I can stick on
the network, but I'm trying to get one built.

I'm also working with somebody in Seattle for file transfer testing.


I think windows should be fine, i just tested it from Vista and FireFox.


Rick Ernst wrote:

Rick. The speedtests are only as good as the hosts they're hosted on and the path by which you reach them.

I use iperf on each end of a link that I'm turning up. I put Linux hosts at both endpoints, but I believe iperf comes in a windows flavor too.


Yup. I use iperf for point-to-point testing, but this is an access
connection which is why I'm looking more for some kind of test host on
Level3 in Seattle rather than a "speed test" site per se.


Hi Rick,

Try an anon ftp or http download from The cluster that serves for has multiple machines with at least single or bonded 10GE interfaces into HEAnet's backbone and then minimum of 10GE on two carriers to the general internet.

Should give you a pretty good speed test. I can max out our GigE links using them for testing.


Paul Kelly
Technical Director
Blacknight Internet Solutions ltd
Hosting, Colocation, Dedicated servers
IP Transit Services
Tel: +353 (0) 59 9183072
Lo-call: 1850 929 929
DDI: +353 (0) 59 9183091


Blacknight Internet Solutions Ltd,
Unit 12A,Barrowside Business Park,
Sleaty Road,

Company No.: 370845


I'm working on turning up our first GigE connection (400mbs CIR) and the
various online speedtests I'm aware of choke after about 100Mbs or so.

Does anybody know of testing sites that can handle higher bandwidth, or
have an ftp host or similar to test against?

I'm connected to Level3, backhauled to Seattle, WA.

You might want to calculate what maximum throughput you can get on one TCP
session (c.f. windowsize and bandwidth delay product when taking into
consideration the RTT between the two endpoints of your test), then start
multiple of these sessions in parallel to fill up your pipe. I strongly
urge you to use a test like netperf/iperf that runs completely from memory
and does not require spinning disks like http/ftp servers usually do.


I turned-up a pair of 10GigE circuits a while back (with a different, unnamed carrier).
They didn't perform too well. When I pushed them for assistance with testing, they revealed that they had multiple IPERF transponders scattered throughout their network.
They were not open to the public, but could be made available for short periods of time (timer-based, requiring repeated re-authorization to use them for an extended period).

It seems likely that Level3 has similar (or superior) testing facilities. A call to some account executives may be required to open the kimono.

Separately, the Super computer centers used to have speed-test servers installed adjacent to their border routers. They were dedicated, tuned hosts specifically for speed testing. One/more of them might be willing to help you out. However, unless one of them happens to use Level3 for commercial transit, your performance will be gated by the excess intervening network(s) and under all circumstances, by the competing traffic on their access circuit.

Finally, I echo the sentiments about avoiding disk I/O.
If you do use FTP download for testing, you may wish to write the local output to the null device. Some ftp clients allow the null device to be specified as the local output file when downloading files. On XP command-line FTP, the device "Nul:" is accepted. On Un*x it is /dev/null. The command-line client on Server 2003 et al does not seem to accept Nul as the local destination file when downloading. (anyone know the correct magic?) Remember to download multiple times; assuming the source server has enough ram, it will cache the file in memory during the first download and successive downloads in rapid succession should be essentially memory-to-memory (if you're using a null device on the receiving end)


Rick Ernst wrote:

If you're turning up a 10 GigE circuit, as a customer I would be asking for
that circuit to be tested with some modern tools such as the JDSU T-BERD.
For the price you're probably paying, it's probably not unreasonable to have
it as part of the turn-up fee.


The attachment circuits physical media was single mode dark fiber mid-span meet. After some tinkering with colo center jumpers, the physical attachment circuits were rock solid. The issue was the internal IP network of the ISP (or lack of same).

You get what you pay for. (At most. Quite often, you get less.)

Frank Bulk wrote:

Thanks to multiple private/public responses.

I was able to get an iperf test and also a close mirror for a DVD iso.

Time to put live traffic on it and see what happens.

Ironically, two days later, I find myself in exactly the same position; needing an iperf box to test a 100Mb client connection against.

In a perfect world, this box would hang off of NYIIX (or Arbinet) and be able to sustain 100Mb of throughput for the duration of a couple of generic iperf tests that I need to perform at a client site in a couple of hours.

Thanks in advance...

I believe there is an ITU standard for testing that could be looked at, but
if you went with the same test gear that SPs use to test their circuits, I
think you would be safe. Hence my mention of JDSU, but I could also add
Agilent (more engineering focused), Anritsu, EXFO, Fluke (more enterprise
focused), and SR Telecom.