Are Orange fiddling speed tests?

Hi,

I suspect that Orange have recently done some jiggery pokery - eg just 'opened up traffic' on speed test ports. Please can any clever people out there look into this, or say if it is possible?

I'm on a LLU exchange -with 8Mb/s service since last August - off-peak max has always been 2.4Mb/s, which invariably dropped to a ridiculous 500kb/s every evening. In the last few days, off-peak downloads have suddenly jumped to 6.3Mb/s according to the thinkbroadband speedtest and broadbandmax.

BUT real-use speed still seems slow and not improved. Using Port 80 on the thinkbroadband test reduces speed to 3.3Mb/s, and local server on speedtest.net returns 2.9Mb/s.

Have Orange done some jiggery pokery - eg just 'opened up traffic' on speed test ports? Please can any clever people out there look into this, or say if it is possible?

I have just referred my case to CISAS, and I get the feeling Orange are going to point to the thinkbroadband.com speedtest for their defence...

What do you mean by "real-use"? Do you mean on peer2peer networks? I've setup a test for you to run.

Download the 1MB file from each of the linked URLs, each is run on a different post. Port 80, 81 and 6881.

80 will go via Orange's cache server. 81 wont, but shouldn't be picked up as p2p. 6881 is the classic BT port, so you'll see if they're (Orange) shaping based on port.

If you see vastly different speeds on the 6881 port then your suspicions are justified, but what do you plan to do? If not, you're just paranoid

See PM for details, this is NOT for public use, do not share these links.

Thank you so for much the help, much appreciated.

I tested the files - they all download at a similar rate. 160KBps off-peak this morning.

16KBps peak last evening though...with a thinkbroadband test of 310kbps...so I guess perhaps Orange are not hiding their blistering speeds after all, lol. And you can half that if I use the Livebox wirelessly - full 310kbps with a wired connection or via another router (only tried and discovered that couple of weeks ago, hmmph...)

Curiously, port 80 no longer slows the test as it did before I posted here.

I have now noticed that since the 'jump' in peak speeds, my IP has changed and the attainable download rate shown on my Livebox increased from 7100 to 8600kbps - I guess this means they really did do something to the line/exchange (useless as it still is at peak times)?

I'd love to know about what different ports are, what shaping is, etc. What's "classic BT"? How does an ISP detect P2P traffic? If you can spare the time, or point to a link!

Thanks again.

I think your jump from 7100 to 8600 will actually be to do with BTs IPStream MAX product. Once it sees that your connection is stable it attempts to move the target noise margin. If you can stay synced with a lower figure for the signal to noise ration, meaning there's more notice, you get higher sync speeds. So that the speed doesn't jump up and down like a jo jo BT's kit will only make changes I think every three days, but now you're at the top of the scale so once Orange roll out some extra capacity you should see much better speeds.

If you're getting the same (slow) speed to that server I linked you there's likely no shaping going on. The server is in Florida so it will be slow. I couldn't set that set up on any UK servers though. Bandwidth in the UK is so expensive.

About the shaping some ISP do.. It used to be that if it was on a peer2peer port each packet would be prioritized lower than say email traffic. If you imagine a router at Oranges offices like a gate where only a cretin about of packets can make it through, basically web and email traffic is allowed to jump the queue and get through the gate..

People fooled that port based system easily so now some ISPs actually inspect the packet's contents. The machine doing this knows what a request for a web page looks like, and knows what the BitTorrent protocol looks like and prioritizes traffic appropriately.

If done correctly QoS or traffic shaping can improve the network for every one. The idea is that the ISP implements the shaping so it only effects people when there is more traffic that the network can bear. IE at peak times.

PlusNet were the first to go this route in the UK, and I was unfortunately one of their customers during that time. At first it was OK but then Plus saw that there was an opportunity for some money making and as the customer base grew they just didn't add any new bandwidth until the point that the traffic shaping was kicking in 24/7 and anything but "gold" traffic was next to useless.

This was after PlusNet's failed "bad boy pipe" experiment. Where all the users that used P2P were assigned to the same pipe and saturated the connection in one directly, making it entirely useless as all times of the day.

As always wikipedia has a lot on the subject:
http://en.wikipedia.org/wiki/Traffic_shaping
http://en.wikipedia.org/wiki/Quality_of_service

Theres a lot of talk in the US right now about this, under the term "net neutrality". Big time backbone ISPs like AT&T want to charge the likes of Google to deliver their content at the same speed, if they don't pay they get shaped to an unspecified speed. Needless to say, only the backbone providers and the people they paid are happy about it..