TrueNet April 2012 report

Forums › New Zealand Broadband › TrueNet April 2012 report

freitasm

BDFL - Memuneh
79270 posts

Uber Geek

Administrator

ID Verified

Trusted

Geekzone

Lifetime subscriber

#102403 15-May-2012 09:12

As we suspected TelstraClear took a big hit. Are they going to recover?

The full report is here.

TelstraClear DSL continues to have very slow evening speeds currently dropping below 80% of their best off-peak speeds. See March report here

During peak time hours (e.g. 8pm) Orcon continues to show the same pattern as last month dropping to 85% of their maximum speed.

amanzi

Amanzi
1292 posts

Uber Geek

ID Verified

Trusted

Lifetime subscriber

#625296 15-May-2012 09:40

I have two questions... how is it possible for TCL cable customers to get up to 18Mbps speed during the day, and, why doesn't TrueNet measure international web page downloads?

JohnButt

374 posts

Ultimate Geek

Trusted

#625300 15-May-2012 09:47

Interesting question on why 18Mb/s, any suggestions? We measure it consistently from lots of probes, although not quite all.

We do measure International, we have not published yet, because we only have one site currently - in Dallas.

NonprayingMantis

6434 posts

Uber Geek

#625305 15-May-2012 09:57

I've read the description, but I still don't really understand the difference between the results for DSL providers vs cable. why index DSL to their top speed, but not do the same for cable?

I appreciate it's hard to compare DSL providers since distance to exchange makes a big difference to actual speed achieved, which is why indexing to top speed achieved is appropriate, but then why single out cable and index the speed differently to DSL?

I mean, lets say that one of the DSL providers was able to consistently achieve >18Mbps all day they might not be too happy with that chart since it implies that TCL cable would be considerably better than them at all times. (indexing the DSL performance to 100% = their maximum actual speed of 18Mbps, but indexing TCL to 100% = far lower than the maximum 15Mbps.

In other words, by this methodology, it is impossible for any DSL provider to ever get above 100% no matter how good their performance is.
All TCL cable has to do to always look better is 'promise' a speed slightly lower than they actually provide, and bingo - they score above 100% every time.

sbiddle

30853 posts

Uber Geek

Retired Mod

Trusted

Biddle Corp

Lifetime subscriber

#625307 15-May-2012 10:04

amanzi: I have two questions... how is it possible for TCL cable customers to get up to 18Mbps speed during the day, and, why doesn't TrueNet measure international web page downloads?

I assume the 25Mbps and 100Mbps users would skew the data?

JohnButt

374 posts

Ultimate Geek

Trusted

#625312 15-May-2012 10:12

@NonprayingMantis good points, but TelstraClear promise 15Mb/s, ie it is their advertised rate and they achieve it all the time as you say.

Whereas DSL suppliers promise as fast as your line can go or similar words, and if they do not have sufficient capacity (for example) then they will not meet that promise. They could offer 80% of the top speed your line can reach, in which case we would have to compare to 80%. (Note: A DSL provider could not deliver 18Mb/s for all customers without selecting which customers they connect based on location.)

ie we are comparing promises, not speeds. The same will be true with Fibre - we will see fibre results in the near future as we gain more fibre volunteers, and the same issue applies there, ie do we measure a 30Mb/s offer as how often they achieve 30Mb/s or the actual speed they achieve? Meeting the promise is more robust.

JohnButt

374 posts

Ultimate Geek

Trusted

#625313 15-May-2012 10:13

sbiddle:
amanzi: I have two questions... how is it possible for TCL cable customers to get up to 18Mbps speed during the day, and, why doesn't TrueNet measure international web page downloads?

I assume the 25Mbps and 100Mbps users would skew the data?

No we are not including them in our stats yet - not enough volunteers - although we may have quite soon

amanzi

Amanzi
1292 posts

Uber Geek

ID Verified

Trusted

Lifetime subscriber

#625326 15-May-2012 10:21

sbiddle: I assume the 25Mbps and 100Mbps users would skew the data?

I assume that when they do start including those customers, they'll need to have separate lines on the graph so they can track the percentage of promised speed.

Tradepub

Free professional, reference and technical white papers (affiliate link).

NonprayingMantis

6434 posts

Uber Geek

#625329 15-May-2012 10:25

JohnButt: @NonprayingMantis good points, but TelstraClear promise 15Mb/s, ie it is their advertised rate and they achieve it all the time as you say.

Whereas DSL suppliers promise as fast as your line can go or similar words, and if they do not have sufficient capacity (for example) then they will not meet that promise. They could offer 80% of the top speed your line can reach, in which case we would have to compare to 80%. (Note: A DSL provider could not deliver 18Mb/s for all customers without selecting which customers they connect based on location.)

ie we are comparing promises, not speeds. The same will be true with Fibre - we will see fibre results in the near future as we gain more fibre volunteers, and the same issue applies there, ie do we measure a 30Mb/s offer as how often they achieve 30Mb/s or the actual speed they achieve? Meeting the promise is more robust.

That is my point really. DSL doesn’t have a ‘promised speed’ and so there is no way to measure it and thus compare providers on their ‘promise’. Basically you are using one methodlogy for all the DSL providers, and a separate one for the cable, which massively skews the results.

Look at it this way. When cable can routinely get above 100% by simply promising a slightly lower speed than they give, and when it is impossible for a DSL provider to EVER get above 100% because you use a different methodlogy so that everytime they improve their performance their '100%' gets adjusted downwards the same amount, then there is something wrong with the methodoogy IMO

Why not use the same method for cable to give a fair comparison? i.e. index to 100% the fastest actual speed achieved, not the ‘promised’ speed, which is meaningless for comparison with DSL.

a separate chart to compare actual speeds achieved would be good, but of course that has weaknesses as a different geographical skew of customers will impact results. e.g. If, say, Vodafone has a very high proportion of urban customers we would see a higher average speed than someone who has a higher proportion of rural cutomers.

amanzi

Amanzi
1292 posts

Uber Geek

ID Verified

Trusted

Lifetime subscriber

#625340 15-May-2012 10:31

NonprayingMantis: Look at it this way. When cable can routinely get above 100% by simply promising a slightly lower speed than they give...

I don't understand how cable can get higher than their advertised speed. TCL cable should never be above 100% as that's the speed that the connection is limited to. And as many cable customers know, there are all sorts of other issues on the TCL network which makes being a cable customer a sh!tty experience (YouTube and iSky buffering, transparent proxy faults, frequent DNS failures, etc...)

NonprayingMantis

6434 posts

Uber Geek

#625454 15-May-2012 13:59

amanzi:
NonprayingMantis: Look at it this way. When cable can routinely get above 100% by simply promising a slightly lower speed than they give...

I don't understand how cable can get higher than their advertised speed. TCL cable should never be above 100% as that's the speed that the connection is limited to. And as many cable customers know, there are all sorts of other issues on the TCL network which makes being a cable customer a sh!tty experience (YouTube and iSky buffering, transparent proxy faults, frequent DNS failures, etc...)

pretty simple really if they wanted to. you would build a service that can do (x)Mbps, but then advertise it as (x-10%)Mbps

for example, the TCL service might be able to go up to (say) 20Mbps, but by advertising it as 15Mbps the truenet report will consistently show them as being greater than 100% performance. A feat that no DSL provider can ever acheive, because the methodology truenet use for DSL is different than what they use for Cable. even if (hypothetically) Snap had all of it's customers on VDSL, getting speeds of >25Mbps, it would still only ever have a peak performance of 100%.

mercutio

1392 posts

Uber Geek

#625480 15-May-2012 14:36

NonprayingMantis:
amanzi:
NonprayingMantis: Look at it this way. When cable can routinely get above 100% by simply promising a slightly lower speed than they give...

I don't understand how cable can get higher than their advertised speed. TCL cable should never be above 100% as that's the speed that the connection is limited to. And as many cable customers know, there are all sorts of other issues on the TCL network which makes being a cable customer a sh!tty experience (YouTube and iSky buffering, transparent proxy faults, frequent DNS failures, etc...)

pretty simple really if they wanted to. you would build a service that can do (x)Mbps, but then advertise it as (x-10%)Mbps

for example, the TCL service might be able to go up to (say) 20Mbps, but by advertising it as 15Mbps the truenet report will consistently show them as being greater than 100% performance. A feat that no DSL provider can ever acheive, because the methodology truenet use for DSL is different than what they use for Cable. even if (hypothetically) Snap had all of it's customers on VDSL, getting speeds of >25Mbps, it would still only ever have a peak performance of 100%.

fwiw vdsl has as an advertised speed of 15megabit, so maybe truenet should be measuring vdsl on that basis to stay consistent. a lot of people get significantly over.

citylink's ethernet network is also going faster than they advertise, as otherwise people complain.