BT Openworld port throttles?

[silver]

well - who knows how valid the tests are, wrt reproducible tests etc.. but.. http://www.adslguide.org.uk/newsarchive.asp?item=393

Users of the Openworld broadband service have variously complained about peer 2 peer application speeds in the last few weeks, mainly affecting Kazza, Gnutella and EDonkey.

An enterprising Software Developer Mark Hutton who is using the BTO Home 500 service has run some tests sending just data between two machines on the various ports that the sharing programs use, a summary of the results follows:

Napster Port 6699 58.1KB/sec
Gnutella Port 6346 3.75KB/sec
Edonkey Port 4662 1.5KB/sec
Kazza 2.8KB/sec
Test (Kazza port+1) 58.1KB/sec

It does very much look like some form of throttling intentional or otherwise is occurring on traffic through BTopenworld. Perhaps BTopenworld are running a maximum bandwidth per port policy, and restricting some ports to ensure that any single application does not overwhelm the network - there is no documentation to suggest this or any feedback to users from the helpdesk. The full post and discussion can be seen in our forums here.

from this ADSL forum post

I've just performed what I deem a definitive P2P test for throttling on certain ports:

I'm using TTCP, a very old *nix utility to measure network throughput - there's a windows port knocking about on the web which I used on my machine.

00:16 2nd October 2001

::TEST 1:: Kazaa port:

Receiver (me, BTo Home 500 ADSL): [wsttcp -r -s -p1214]
Sender (XMbit conn to Nildram backbone): [ttcp -t -s -p 1214 <myhost.com>]

I sent raw data for 60 seconds:

wsttcp-r: 628432 bytes in 217.47 real sec = 2.82KB/sec (23118.04 bps)
wsttcp-r: 315 I/O calls, msec/call = 706.95, calls/sec = 1.45628432 1001978074.78 1001978292.25 217.47 23118.04

NB: The output on screen of the raw data being received was extremely bursty and continued far beyond the 60 seconds of sending, as shown above (217.47 secs)


::TEST 2:: Kazza port + 1:

Receiver (me, BTo Home 500 ADSL): [wsttcp -r -s -p1215]
Sender (XMbit conn to Nildram backbone): [ttcp -t -s -p 1215 <myhost.com>]

I sent raw data for 60 seconds:

wsttcp-r: 3646064 bytes in 61.38 real sec = 58.01 KB/sec (475250.70 bps)
wsttcp-r: 823 I/O calls, msec/call = 76.36, calls/sec = 13.413646064 1001978496.88 1001978558.25 61.38 475250.70

NB: Extremely fast output on screen, stopped as soon as the source was stopped.

v interesting if true, happy to test it out by port testing between me an somone on BTO to see what happens

Sil