[v6test] Test download server

[Iljitsch van Beijnum]

On 23 mrt 2008, at 13:14, Kevin Day wrote:

> If you're getting really bad or really good download speeds off that,
> let me know off list.

So middle of the road speeds on-list?

  8  testdownload.ipv6experiment.com (216.14.100.2)  100.405 ms   
100.341 ms  100.423 ms
# ftp -4 http://testdownload.ipv6experiment.com/24-h264-aac.mp4
36228544 bytes retrieved in 00:57 (614.64 KB/s)

  8  unassigned.v6.your.org  131.268 ms  130.759 ms  131.747 ms
# ftp -6 http://testdownload.ipv6experiment.com/24-h264-aac.mp4
36228544 bytes retrieved in 01:20 (439.18 KB/s)

The difference in performance seems to correlate rather nicely with  
the difference in RTT (why don't Teleglobe and HE peer in Amsterdam??)  
and indeed:

18:00:43.941323 sequoia.muada.com.59412 > unassigned.v6.your.org.http:  
S 3706957527:3706957527(0) win 65535 <mss  
1440,nop,nop,sackOK,nop,wscale 1,nop,nop,timestamp 61301765[|tcp]>  
[flowlabel 0x94ab2]
18:00:44.072302 unassigned.v6.your.org.http > sequoia.muada.com.59412:  
S 3653372740:3653372740(0) ack 3706957528 win 16384 <mss  
1390,nop,wscale 5,sackOK,timestamp 1671422917 61301765> [flowlabel  
0x142bf]

Strange that my box isn't using a larger window scale option, but the  
one at your end should be sufficient. What's with the low MSS, though?