VDSL is delivering for NZ!

Forums › New Zealand Broadband › VDSL is delivering for NZ!

JohnButt

374 posts

Ultimate Geek

Trusted

#150264 16-Jul-2014 20:03

TrueNet published the June Urban Broadband Report yesterday, the results include an interesting comparison with our first Australian ISP with enough ADSL probes to compare:

Australia Internet Speed Falls Well Behind: NZ 9 v Aussie 3

New Zealand median speed for ADSL is 9Mb/s but for Australia it was just 3Mb/s in TrueNet's June results.

Copper services in NZ include VDSL over FTTN, so the difference is much greater. A substantial 2% growth in fibre probes occurred in June with 9 probes upgrading from DSL.

TrueNet now has 420 probes with 74 of those on UFB fibre and 90 on VDSL, nearly all volunteer upgrades from ADSL. The premium UFB fibre product continues to fall below the advertised speed of 100Mb/s for most users.

Link to report here.

"Australian ISPs do not have the "luxury" of cabinets, or VDSL, or FTTN, so speeds are A LOT slower on average than NZ. To show the difference, Chart 7 compares the distribution of tests on ADSL between NZ and Australia, as well as VDSL in NZ.

Results from our Australian test probes show speeds predominantly in the 2-5Mb/s range with NZ test probes more evenly distributed in the the 6-14 Mb/s range. "

Did you know that VDSL was 4% last Chorus report and at the growth rate then had it should be about 8% now!

raytaylor

3835 posts

Uber Geek

Trusted

#1090297 17-Jul-2014 01:14

This should be a sticky post so i can refer back to it during an argument

Ray Taylor

There is no place like localhost

Spreadsheet for Comparing Electricity Plans Here

Mighty Ape

Shop Mighty Ape for electronics, games, computers books and more (affiliate link).

ripdog

548 posts

Ultimate Geek
Inactive user

#1090304 17-Jul-2014 04:33

Thanks for your hard work as always, John's Butt! And of course all the other folks at TrueNet. Always feels nice to have a bit of smug superiority over the aussies, after they stole half our population in the mining exodus. These results are hardly a surprise, from what I've heard from aussies on reddit, though.

Do you have any NBN probes yet?

Bit sad to see my ISP Orcon still trailing snap in download speeds (specifically 100mbit) and national latency. Good work on international latency and webpage downloads, though!

Oh yeah, are you interested in a 100mbit Orcon probe in Wanganui? I think you already have plenty of them, but it's worth asking...

JohnButt

374 posts

Ultimate Geek

Trusted

#1090447 17-Jul-2014 09:52

TrueNet funding is limited to DSL and Cable apart from where an ISP funds some probes (Orcon, Snap & Telecom Fibre probes). That means we have funds to accept 20 Telecom Fibre volunteers, but we do not have funding for other ISPs at present.

The download speed issue for fibre is complicated, I am working on a better method of presenting the data. Basically the "average" or median is not a great way to represent the data. On fibre, users typically get the speed expected - for 100Mb/s that is 94Mb/s, although should be 100Mb/s - or get something worse. A small number get 100Mb/s. The average or median usually pops in below the expected speed, messing with the appearance in a chart like the one we have published. Some probes measure very much lower than 94, often as low as 20-25Mb/s as below.

Here is a sample of my thinking for comment, i.e. would this work for a comparison of ISPs? I put the chart together to study the spread but it looks a good chart for comparing LFCs and ISPs. Note this is actual tests, so reflects the number of probes by LFC - i.e. Northpower = 1 The number of tests however, is not a good representation of probes, some test every hour others every 5 hours. We have 74 Fibre probes with a couple of Telecom on the way.

ripdog

548 posts

Ultimate Geek
Inactive user

#1090492 17-Jul-2014 10:13

That's interesting stuff, thanks for the info.

It's extremely disappointing that (government, I presume?) universal funding is only provided for DSL and Cable. Accepting funding from ISPs is a pretty big conflict of interest. I'm sure you guys do the best you can, but it's still not a fantastic situation.

LFC comparison is a very good point - with regional monopolies granted by regulations, it is extremely important that the LFCs are named and shamed if they fall behind. I can't switch LFC like I can ISP, after all!

As for how to present the data, I think a horizontal bar graph with one bar per LFC, each split into coloured segments for speed segments. It may also be worthwhile to clump the "good enough" speeds from, say. 80-100 to draw attention to the unusually low speeds. I think this would allow easier side-by-side comparison than the chart you present earlier. Due to the size of Chorus, the blue takes over the chart and I honestly don't know which LFC wins relatively.

These are national tests, right?

I'm quite disturbed by the number of below-80 results on there, though. Current 100mbit plans mean that there should be 0 contention at the OLT level, so I guess this is an ISP problem? Would be interested in seeing the same graph with ISP data instead.

JohnButt

374 posts

Ultimate Geek

Trusted

#1090514 17-Jul-2014 10:34

OK So I can meet that challenge :-)

You may be right, the bars look quite good, they definitely show the nature of the problem, in all categories. The issue is that the light blue boxes are so small!

Is 80-90 "Good enough"? for that matter is 90-100 good enough if it is sold as 100?

sbiddle

30853 posts

Uber Geek

Retired Mod

Trusted

Biddle Corp

Lifetime subscriber

#1090525 17-Jul-2014 10:37

How are you planning on factoring in the SLAs on EIR traffic as part of right performing?

JohnButt

374 posts

Ultimate Geek

Trusted

#1090534 17-Jul-2014 10:49

What do you suggest?

sbiddle

30853 posts

Uber Geek

Retired Mod

Trusted

Biddle Corp

Lifetime subscriber

#1090542 17-Jul-2014 11:04

JohnButt: What do you suggest?

'

I don't have one - the reality is as speeds get faster, measuring speed as a % of maximum throughput becomes an increasingly impossible way to measure performance.

Behodar

9246 posts

Uber Geek

Trusted

Lifetime subscriber

#1090546 17-Jul-2014 11:11

A little off-topic, but https://www.truenet.co.nz/volunteer-details (linked from https://www.truenet.co.nz/get-involved-become-volunteer-tester) gives Access Denied.

BMarquis

374 posts

Ultimate Geek

Trusted

Chorus

Lifetime subscriber

#1090565 17-Jul-2014 11:29

I still hate 90-100 being below spec. :)

In reality it is well within the specification set on the standard UFB offers set by the TCF and CFH - which were Layer 2 rates.
Of course that changes with the new 'Accelerate' offers, so You'll need to differentiate 'Accelerate' and 'Reference' offers soon.

As for not measuring a % of max/peak throughput - I agree. You'll want to report along the same lines as the Accelerate SLAs. I suspect reporting on the committed Service SLAs would be a minimum subset of what you would report on with other interesting facts, etc on top of that.

jonb

1739 posts

Uber Geek

Trusted

#1090568 17-Jul-2014 11:41

I would guess the UFB probes measuring 20-25Mb/s are actually be on a 30Mb/s plan?

JohnButt

374 posts

Ultimate Geek

Trusted

#1090632 17-Jul-2014 12:35

"A little off-topic, but https://www.truenet.co.nz/volunteer-details (linked from https://www.truenet.co.nz/get-involved-become-volunteer-tester) gives Access Denied." Fixed thanks a lot - I had a warning yesterday, but could not find the link causing it

"the reality is as speeds get faster, measuring speed as a % of maximum throughput becomes an increasingly impossible way to measure performance" I struggle with this, I think you mean that as speeds get faster the variation may get too small to measure. That will become obvious if it is true, and I will simply stop publishing - as I did this month with the relatively useless fixed website testpages.
However, currently the issue is the provision of the speed purchased, or even close to it.

The spec may be layer 2 rates, but the product is sold layer 3 and on 100Mb/s, in fact CFH say that is the minimum! So I am comparing performance to advertised spec:
Snap website = "Offering speeds of up to 100Mbps download and up to 50Mbps upload"
Orcon website = "Fibre 100 plans offer you super-charged download speeds of up to a massive 100Mbps"
CFH website = "Ultra-Fast Broadband is taken to mean the availability of broadband services at a minimum speed of 100 Mbps Downstream (from the Internet to the user) and a minimum of 50 Mbps Upstream (from user to the Internet)."

The "Accelerate SLA" is a marketing branding from CFH to ISPs, ultimately this product range is just a speed that will be advertised by the ISPs which we will then measure against. Note that Chorus commented in February that they would increase layer 2 to enable layer 3 to reach full speed and Enable say they made layer 2 106Mb/s in December last year just after we published, yet many probes on these connections are still well below 100Mb/s

"I would guess the UFB probes measuring 20-25Mb/s are actually be on a 30Mb/s plan?" So do I, but I have seen the volunteer's invoice and it is 100Mb/s being billed. We have had, I think, three volunteers sit on 30Mb/s for some time while the ISP recognises that they have allocated the wrong speed. I wonder how big the error rate is considering the size of our sample.

Lazarui

136 posts

Master Geek

#1093907 22-Jul-2014 20:48

Psst. John,

VDSL is available to over 80% of NZ homes and businesses from FTTN Cabinets, or from the exchange. The planned rollout of VDSL over FTTN in Australia may be arriving too little too late

Is wrong, It's not VDSL that is available to over 80% its 'high speed broadband' which includes ADSL2+.

;)