"Don't blame us, blame your speed test"

Originally Posted by rossdagley

http://www.speedtest.bbmax.co.uk/res...1835&v=5411395

Both of those are "acceptable" sites as far as I can see. Tested via my Windows 2003 box co-located in interXion, London.

Tested at home, on my VM 20mbit connection:

http://www.speedtest.bbmax.co.uk/res...2090&v=5411423 (10mbit)



Gee - I'm sure pleased I'm paying £36 for this 10mbit quality service. :|

Post your modem stats here please 192.168.100.1 username and password is root root

also can you post what UBR your connected to

[rossdagley]

Originally Posted by darthlinux

Post your modem stats here please 192.168.100.1 username and password is root root

also can you post what UBR your connected to

Network Access :  Allowed 
Maximum Downstream Data Rate :  20480000 
Maximum Upstream Data Rate :  768000 
Maximum Upstream Channel Burst :  1600 

Downstream Lock :  Locked 
Downstream Channel Id :  7 
Downstream Frequency :  331000000 Hz 
Downstream Modulation :  QAM256 
Downstream Symbol Rate :  5360.537 Ksym/sec 
Downstream Interleave Depth :  taps32Increment4 
Downstream Receive Power Level :  -8.8 dBmV 
Downstream SNR :  36.8 dB 

Upstream Lock :  Locked 
Upstream Channel ID :  3 
Upstream Frequency :  34000000 Hz 
Upstream Modulation :  QAM16 
Upstream Symbol Rate :  2560 Ksym/sec 
Upstream transmit Power Level :  44.0 dBmV 
Upstream Mini-Slot Size :  2 

Boot Code Version : 3.1.6d 
Software Version : 2.94.1014 
Hardware Version : 1.19

Not that it'll help - I do ocasionally get 20mbit - occasionally being the operative word. It's oversubscription. Combined with STM, it's just not worth the hassle. Hence I'm leaving.

Edit to add connection: cpc1-hatf2-0-0-cust83.lutn.cable.ntl.com

[dev]

to me, 10mb on 20mb wouldnt mean oversubscription, it would be if they sold it with a contention ratio of anything better than 2:1

[Fingers]

Originally Posted by dev

to me, 10mb on 20mb wouldnt mean oversubscription, it would be if they sold it with a contention ratio of anything better than 2:1

They avoid mentioning any contention ratios.

Back when I first got cable in 1998, Cable London promised a contention ratio of 20:1 in their terms and conditions. It was used as a selling point because BT ADSL (the only type available then - pre LLU) was a choice between 25:1 (premuim cost) or 50:1.

However, mentioning contention ratios is deliberately avoided now - shame, because it would be an easy way of nailing ISP's to the wall for failing to keep to their side of the contract. They just simply keep piling more and more subscribers onto the UBR & POP's. I am currently getting slower speeds on 20mbit cable than I used to get on ISDN!

[dev]

Originally Posted by Fingers

They avoid mentioning any contention ratios.

Back when I first got cable in 1998, Cable London promised a contention ratio of 20:1 in their terms and conditions. It was used as a selling point because BT ADSL (the only type available then - pre LLU) was a choice between 25:1 (premuim cost) or 50:1.

However, mentioning contention ratios is deliberately avoided now - shame, because it would be an easy way of nailing ISP's to the wall for failing to keep to their side of the contract. They just simply keep piling more and more subscribers onto the UBR & POP's. I am currently getting slower speeds on 20mbit cable than I used to get on ISDN!

they probably avoid saying it because it's less relevant now, having 20x20mbit customers fighting over 20mbit of bandwidth would be fine for people who don't use it a lot, but 20 of the high downloaders and it looks worse

[Axegrinder]

Originally Posted by rossdagley

Network Access :  Allowed 
Maximum Downstream Data Rate :  20480000 
Maximum Upstream Data Rate :  768000 
Maximum Upstream Channel Burst :  1600 

Downstream Lock :  Locked 
Downstream Channel Id :  7 
Downstream Frequency :  331000000 Hz 
Downstream Modulation :  QAM256 
Downstream Symbol Rate :  5360.537 Ksym/sec 
Downstream Interleave Depth :  taps32Increment4 
Downstream Receive Power Level :  -8.8 dBmV 
Downstream SNR :  36.8 dB 

Upstream Lock :  Locked 
Upstream Channel ID :  3 
Upstream Frequency :  34000000 Hz 
Upstream Modulation :  QAM16 
Upstream Symbol Rate :  2560 Ksym/sec 
Upstream transmit Power Level :  44.0 dBmV 
Upstream Mini-Slot Size :  2 

Boot Code Version : 3.1.6d 
Software Version : 2.94.1014 
Hardware Version : 1.19

Not that it'll help - I do ocasionally get 20mbit - occasionally being the operative word. It's oversubscription. Combined with STM, it's just not worth the hassle. Hence I'm leaving.

Edit to add connection: cpc1-hatf2-0-0-cust83.lutn.cable.ntl.com

Your downstream power is a little low for an ex ntl connection.

[rossdagley]

Like I said - I occasionally get 20mbit. Just not today. Or yesterday. Or for more than an hour during 10am-3pm, or 4pm till 11pm...

I still find the test here is the most consistant ;

Today I get this ;

Fri, 31 Oct 2008 13:11:05 UTC

Test 1: 1024K took 453 ms = 2260.5 KB/sec, approx 18627 Kbps, 18.19 Mbps
Test 2: 1024K took 437 ms = 2343.2 KB/sec, approx 19308 Kbps, 18.86 Mbps
Test 3: 1024K took 438 ms = 2337.9 KB/sec, approx 19264 Kbps, 18.81 Mbps
Test 4: 2048K took 890 ms = 2301.1 KB/sec, approx 18961 Kbps, 18.52 Mbps

Overall Average Speed = approx 19040 Kbps, 18.6 Mbps

[Ignitionnet]

Originally Posted by dev

to me, 10mb on 20mb wouldnt mean oversubscription, it would be if they sold it with a contention ratio of anything better than 2:1

Broadband doesn't work on contention ratios but visible contention. In a lightly used area 200:1 might be fine, in a heavily utilised area 20:1 may not be good enough.

Areas are considered oversubscribed when contention becomes visible. No contention ratio is given as it would likely scare people

[Ignitionnet]

Was going to post some speedtests but I appear to be getting a bit of congestion this evening.

Download Speed: 9011 kbps (1126.4 KB/sec )
Upload Speed: 708 kbps (88.5 KB/sec )

[rossdagley]

Ditto.

I was STM'd from about 17:00 this evening - since about 22:00 I'v ebeen getting 1.4Mbit/sec - nothing more. I used to get 2+Mbit. Congestion/oversubscription etc.

[Sambora]

Back on topic If you look at the detail of the Press Release that VirginMedia made you'll see that they were saying some tests are not as good as others. If Virgin kept their testing environment consistent then this is pretty good insight. Reference this graph;



where you can see several sites (not named) were close at the top, and the rest ranged from not so good to downright bad.

The slide also referes to headline speed so perhaps these results were 20Meg rather than 50Meg?

The whole article is at Hexus.channel

The article also mentions monthly testing so maybe we will see some more in the next couple of weeks.

I'd guess www.thinkbroadband may be predicting some good results so have got their piece in early if they are confident!

[Ignitionnet]

My issue with said article is this though:

The only accurate way to test access speed on 50Mb at the moment is to use a speedtest embedded within the Virgin Media network - this eliminates a number of issues - payload size, routing and internet congestion, bandwidth costs, and gives a truer picture of actual throughput.

Who cares what the access speed is though? What does it prove if 50Mbit is available inside the Virgin Media network if it's not reachable as soon as you step outside it?

IMHO while it may give a truer picture of your connection to the CMTS it does not give a picture of 'actual throughput' in practise. If there were issues due to routing, bandwidth costs, etc, I'd much rather find them out through a speedtest located at LINX than be constantly told all is fine as I can achieve good performance inside my SP's network.

Thinkbroadband's article demonstrates that they are able to serve 50Mbit to any well connected network, and realistically there is no reason why any ISP in the UK cannot at least provide 50Mbit from their core network to TB or any other site that has sufficient bandwidth - this is what LINX is for.

While it's useful to have an internal speedtest to eliminate issues closer to home I will continue to go offnet to measure my performance. After all, the internet is not hosted on the Virgin Media network, it's merely a conduit to the internet.

Something from these chaps http://www.ookla.com/ on the VM network would be ideal to assist with resolution of access / transport speed issues though!

[Gopher]

Originally Posted by moaningmags

Totally off topic, but I met my husband through CB radio

Our CB is now in the loft but our Antron 99 is still standing poker straight on the roof.

Still use it when travelling down south with modulator on the car, any trucker who sees it on your car will let you in or out etc etc.

Back on topic, I've yet to have a speed test give me an accurate result, I rely on my download speeds for that.

10.33 for a corkscrew

[Sambora]

Originally Posted by Broadbandings

A very poor speed tester that consistently gives bad results on anything apart from low bandwidth links and should be avoided is, amusingly, this one:

http://www.broadbandchoices.co.uk/speed-test.asp

For all their metrics, etc, I have never been able to break around 6Mbit on their tester across two ISPs, while others give a good result.

Originally Posted by Broadbandings

Something from these chaps http://www.ookla.com/ on the VM network would be ideal to assist with resolution of access / transport speed issues though!

Just checked Broadbandchoices out and you've made an interesting / confusing point

Their speedtest is

from these chaps http://www.ookla.com/

Attachment 17165

It also clearly offers the user a chance to choose "Up to 50Mb" through a drop down, yet it seems from what you say that is the absolute max their host can offer?

Seems like this is (again) the point that VM are making? Why would a site purport to be able to test up to 50Meg when it clearly cannot support that rate?