Your broadband speeds

Forums › New Zealand Broadband › Your broadband speeds

1 | 2 | 3 | 4 | 5

2320 posts

Uber Geek

Trusted

Lifetime subscriber

#585353 22-Feb-2012 20:48

@JohnButt
How does traffick shaping affect you? Can you measure it?

PS I've had the Truenet box connected for nearly a year now & have not noticed any impact. I'll keep my unlimited cap for as long as possible. Go ahead with your tests. ;-)

Please keep this GZ community vibrant by contributing in a constructive & respectful manner.

JohnButt

374 posts

Ultimate Geek

Trusted

#585365 22-Feb-2012 21:25

re "How does traffick shaping affect you? Can you measure it?"

tough answering this question, it could have two meanings, so I hope I get the right one.

Our probes will simply measure the impact of traffic shaping, which is likely to be the cause of the dip in speeds during peak periods, but we do not measure if it exists yet.

M-labs have developed a methodology to test for traffic shaping and we may implement that at some stage, though it is not on our urgent radar yet.

Thanks for volunteering and offering us plenty of capacity, I presume you volunteered 5GB/month so we would have used your probe to do the banks study :-)

kyhwana2

2566 posts

Uber Geek

#585369 22-Feb-2012 21:27

ohoh..
Noise Margin: 11.0 dB 12.0 dB
Output Power: 18.5 dBm 12.5 dBm
Attenuation: 24.0 dB 9.0 dB
Speed (kbps): 0 17179 0 868

sonyxperiageek

2958 posts

Uber Geek

Trusted

#586786 25-Feb-2012 20:42

Slingshot $40 50g plan

Mode:ADSL2+ Line Coding:Trellis On Status:Connected Link Power State:L0 DownstreamUpstreamSNR Margin (dB):6.7 5.7 Attenuation (dB):36.5 20.5 Power (dBm):0.0 12.8 Attainable Rate (Kbps):10536 1116 Rate (Kbps):8844 1121 MSGc (number of bytes in overhead channel message):52 36 B (number of bytes in Mux Data Frame):61 59 M (number of Mux Data Frames in FEC Data Frame):4 1 T (Mux Data Frames over sync bytes):5 1 R (number of check bytes in FEC Data Frame):6 0 S (ratio of FEC over PMD Data Frame length):0.8944 1.6842 L (number of bits in PMD Data Frame):2272 285 D (interleaver depth):16 1 Delay (msec):3 0
Edit: added statistics

Sony

Kixtro

49 posts

Geek

#596930 18-Mar-2012 22:34

Telecom pro $115 60GB plan.

gumdigger

429 posts

Ultimate Geek

#596953 19-Mar-2012 01:32

Telecom

blackjack17

1705 posts

Uber Geek

#596960 19-Mar-2012 06:37

Mt Roskill

Vodafone

Electric kiwi signup credit Sharesies credit

Quic Broadband

Move to New Zealand's best fibre broadband service (affiliate link). Note that to use Quic Broadband you must be comfortable with configuring your own router.

BobbyNZ

1 post

Wannabe Geek

#597338 19-Mar-2012 19:35

Slingshot Unlimited for $60

moxpearl

123 posts

Master Geek

#597671 20-Mar-2012 13:38

Noise Margin 12 dB
Attenuation 7 dB

:/

cyril7

9058 posts

Uber Geek

ID Verified

Trusted

Subscriber

#597687 20-Mar-2012 14:00

Hi, Moxpearl something not right there, whats the sync rate, just edit it to the above post

Cyril

Nakedcity

93 posts

Master Geek

#597722 20-Mar-2012 14:43

Slingshot Broadband 150 GB/month

NonprayingMantis

6434 posts

Uber Geek

#597726 20-Mar-2012 14:54

JohnButt: Truenet has selected volunteers to ensure we have a common size sample for each ISP rather than a representative sample of NZ. We did this because we are more interested in comparing NZ ISPs but less interested in international comparisons. The lack of a need to compare NZ results means we do not need a statistically valid sample for NZ.
However, our sample is certainly big and the collection process robust, we can only speculate on how representative it is. We have details of distances from the DSLAM which we will at some stage compare with NZ cable lengths to see if the selection is random, but there is no reason why it should not be, most volunteers are unlikely to choose to volunteer with any bias in relation to their copper length.

I suspect that might be incorrect, not for the overall testing, but rather for the ISP comparison.

Some ISPs will have a much higher proportion of customers on long loop lengths than other ISPs, and so will naturally have lower average speeds all else being equal.

Take Telecom, for example. They are known to have a disproportionatly high share of non-urban customers compared to any of the ISPs which have unbundled.

Unbundling tends to only happen in large exchanges where customers tend to be closer to the exchange than the smaller exchanges where distances will likely be greater.

So if your sampling is random, then it will have a higher proportion of 'long loop length' people who are Telecom customers compared to the ISPs which have unbundled, and a higher relative proportion of 'short loop length' people who re customers of Telstraclear, Vodafone, and Orcon.

Will be interesting to see if this is true or not,

mercutio

1392 posts

Uber Geek

#597758 20-Mar-2012 15:33

NonprayingMantis:
JohnButt: Truenet has selected volunteers to ensure we have a common size sample for each ISP rather than a representative sample of NZ. We did this because we are more interested in comparing NZ ISPs but less interested in international comparisons. The lack of a need to compare NZ results means we do not need a statistically valid sample for NZ.
However, our sample is certainly big and the collection process robust, we can only speculate on how representative it is. We have details of distances from the DSLAM which we will at some stage compare with NZ cable lengths to see if the selection is random, but there is no reason why it should not be, most volunteers are unlikely to choose to volunteer with any bias in relation to their copper length.

I suspect that might be incorrect, not for the overall testing, but rather for the ISP comparison.

Some ISPs will have a much higher proportion of customers on long loop lengths than other ISPs, and so will naturally have lower average speeds all else being equal.

Take Telecom, for example. They are known to have a disproportionatly high share of non-urban customers compared to any of the ISPs which have unbundled.

Unbundling tends to only happen in large exchanges where customers tend to be closer to the exchange than the smaller exchanges where distances will likely be greater.

So if your sampling is random, then it will have a higher proportion of 'long loop length' people who are Telecom customers compared to the ISPs which have unbundled, and a higher relative proportion of 'short loop length' people who re customers of Telstraclear, Vodafone, and Orcon.

Will be interesting to see if this is true or not,

i'd say that is wrong. cabinet's tend to be close, unbundling is often 2km+. the thing is, exchanges often aren't very close to residential users. cabinets on the other hand are. the closest exchange to me is something like 7km away, in a neighbouring suburb. but the closest cabinet is around 2.1km away.

unbundling can happen on the exchange, but not the cabinet. so if my line was unbundled i'd have a much reduced speed.

NonprayingMantis

6434 posts

Uber Geek

#597763 20-Mar-2012 15:49

mercutio:
NonprayingMantis:
JohnButt: Truenet has selected volunteers to ensure we have a common size sample for each ISP rather than a representative sample of NZ. We did this because we are more interested in comparing NZ ISPs but less interested in international comparisons. The lack of a need to compare NZ results means we do not need a statistically valid sample for NZ.
However, our sample is certainly big and the collection process robust, we can only speculate on how representative it is. We have details of distances from the DSLAM which we will at some stage compare with NZ cable lengths to see if the selection is random, but there is no reason why it should not be, most volunteers are unlikely to choose to volunteer with any bias in relation to their copper length.

I suspect that might be incorrect, not for the overall testing, but rather for the ISP comparison.

Some ISPs will have a much higher proportion of customers on long loop lengths than other ISPs, and so will naturally have lower average speeds all else being equal.

Take Telecom, for example. They are known to have a disproportionatly high share of non-urban customers compared to any of the ISPs which have unbundled.

Unbundling tends to only happen in large exchanges where customers tend to be closer to the exchange than the smaller exchanges where distances will likely be greater.

So if your sampling is random, then it will have a higher proportion of 'long loop length' people who are Telecom customers compared to the ISPs which have unbundled, and a higher relative proportion of 'short loop length' people who re customers of Telstraclear, Vodafone, and Orcon.

Will be interesting to see if this is true or not,

i'd say that is wrong. cabinet's tend to be close, unbundling is often 2km+. the thing is, exchanges often aren't very close to residential users. cabinets on the other hand are. the closest exchange to me is something like 7km away, in a neighbouring suburb. but the closest cabinet is around 2.1km away.

unbundling can happen on the exchange, but not the cabinet. so if my line was unbundled i'd have a much reduced speed.

I'm comparing rural exchange customers (who have longer loop length and more likley to be Telecom) with urban exchange customers (who have shorter loop length and more likley to be Voda./TCL/Orcon etC) customers), not cabinet fed customers with exchange fed customers

But you may be right, I don't know.
But don't Chorus have to provide a service to unbundled ISPs that allows them to sell their service to cabinet customers (via chorus wholesaled broadband from the cabinet and the ISPs voice equipment in the exchange)?

moxpearl

123 posts

Master Geek

#597773 20-Mar-2012 16:03

cyril7: Hi, Moxpearl something not right there, whats the sync rate, just edit it to the above post

Cyril

Heya would not let me edit it

Data Rate:
Upstream 809 Kbps.
Downstream 9306 Kbps

1 | 2 | 3 | 4 | 5