I retired from personal blogging in July 2008 but you can find me over at blog.xero.com
Follow @roddrury
Posted by rod@drury.net.nz in Communications at 7:37 pm on Monday, 12 March 2007
Hawkes Bay ISP Airnet is currently New Zealand’s fastest ISP.
http://www.speedtest.net/global.php?random=73877&continent=6&country=5
Good one Sam, Cecil, Ben and the team.
I was asked last night what is the the impact of some of the technologies we’re seeing. Maybe it’s because I’m a dad but one impact that I’m seeing is a resurgence of the provinces. Our first Hawkes Bay Xero, Louise, started last month as part of our test team working remotely over an Airnet connection.
blog
Also interesting: click on the Australian figures, especially the upload ones. Quite a difference compared to the NZ measurements.
Airnet was my first choice for a ‘net connection when I moved to Napier (based purely on your recommendation)… but their pricing , for the sort of monthly traffic volume I regularly use, just didn’t come close to being competitive - around four times the price, for 1/5th of the traffic.
While download speeds are interesting, personally, I’m far more interested in a lower latency connection. I regularly work on (Linux) servers that are in the US and Finland, over ssh. I’d deal with slower up/downloads if the latency was significantly better.
I just did a speed test from my home connection, and this raised two interesting points:
1. The speed test is ’sponsored by Airnet’ and there’s a server in Napier. Does that mean that the server I was testing against was physically on Airnet’s network?
2.
You’d think that I, of all people, would know what could and could not be posted in these comments :)
item 2. above is supposed to be a link to : http://www.speedtest.net/result/98193246.png
Must be time for coffee
Great to see you guys encouraging hiring people remotely. There are 100,000 Aucklanders sitting in traffic right now who could be working remotely for companies from whereever instead, including all the regions. It’s time more companies took advantage of this in New Zealand.
The Canadians are reportedly rolling out a 100 mps broadband service over the next few months…now that is what I call unleashed.
Rod, not sure these graphs have a lot of meaning expecially the second one. You cannot tell me that Clevedon which is way out in the country is the 6th fastest so called city; more like a village.
I am not interested in who is the fastest ISP but how fast my Internet connection works.
In saying this, if Hawkes Bay can get a decent service then everywhere should be able to.
As for the 100,000 Aucklanders sitting in traffic at 9:30 I am not sure those figures are accurate. Not sure from personal experiences that remote working always produces the highest productivity.
We (Epitiro Technologies) are launching our broadband customer experience measurement and benchmarking services in NZ next quarter. We’ve measuring the UK market for quite a few years (during the LLU process)and hope to be doing the same here. It’ll be interesting to see if, how, and how fast the market improves from a customer’s perspective.
Bad statistics experiments annoy me. To the level of ranting. :)
Let’s debunk these.
1) Speedtest appears to be running a server on Airnet’s network (thanks Andrew!)
2) There is no information on the number of samples, nor of the uniqueness
of the sample sources.
3) Speedtest doesn’t break down the samples by interconnect type, only by source
ISP. This means that modem connections are grouped with broadband
connections.
All of which makes it very, very easy to game the test. The local server indicates that they are only measuring the performance to the head end of the WiMax network, not their national, let alone international, interconnects.
It also means that any other carrier testing in New Zealand is likely to also use AirNet’s server, giving AirNet a distinct home field advantage.
The unspecified number of samples could indicate that it is a controlled device, close to the tower. Which is odd, since WiMax can supposedly go a lot faster. ;)
The unspecified number of samples also doesn’t give us any indication of spread of samples, allowing the tester to choose a low traffic period to make the test.
Which makes it a great piece of marketing, but terrible information on which to make a buying decision. It’s certainly gotten everyone talking about AirNet! Does anyone remember when the video card manufacturers were caught putting special case tests into their drivers for the performance test suites? This leaves that sort of taste in my mouth.
I am very sure that if Telecom hosted speedtest.net servers next to their DSLAMs, a lot of people would be reporting 7+mbps, and users would instantly throw out the data because it didn’t reflect actual experience.
If all people want to do is check their local line speed, there are easy ways to do it. For example, I’m on TCL’s cable. I regularily use their debian mirror as a source of large files to test my transfer rates.
Here’s a good one:
http://debian.paradise.net.nz/debian-cd/3.1r0/debian-31r0a-i386-netinst.iso
I am sure that Telecom also has some local mirrors of large files. Look for mirrors of computer game patches.
For national interconnect tests, you can always try other mirrors, off your ISP’s network. You can always try:
http://ftp.nz.debian.org/iso/Debian_3.0_r2/debian-30r2-i386-binary-1_NONUS.iso
which is hosted by citylink.
International? I like mirror.aarnet.edu.au.
Jason
Yup, Airnet’s servers reported a (not bad) 4.7Mbps up/down from my NOC; whereas my direct upstream provider reports a (more realistic) 8.9Mbps up/down (transit that I pay for and am very happy with).
Speed tests to/from your own NOC are fine for customer satisfaction that the local loop is fine. It’s up to the ISP to ensure that their upstream is engineered to adequate levels of contention, latency and loss. (No worse than 32:1 contention, no worse than 30ms national latency for example).
Since everyone is in the same boat, that ISP engineering should be starting to tend towards uniform. That leaves us the local loop to engineer to adequate levels of customer satisfaction.
Is it easier to do that on:
- licensed spectrum? (WiMax)
- public spectrum (2.4, 5.3 and 5.8GHz for example)
- ADSL to central exchanges (up to 6km from customer endpoint)?
- ADSL to distributed DLSAMs (within 250m of customer endpoint)?
- fibre to the customer endpoint?
Jason
Airnet do not operate a WiMax network, rather they operate in the public spectrum (5.8Ghz)
Odd. I was pretty sure that I had read somewhere that they were (while researching my post).
Yep, here we go:
http://www.wimax-industry.com/ar/3t.htm
http://www.scoop.co.nz/stories/SC0512/S00045.htm
Have they not completed roll-out? Or have they decided to go in a different direction?
Unsure, but know for sure its 5.8Ghz gear that they are currently using.
It looks like 5.8Ghz is a wimax band now?
http://www.wimaxforum.org/technology/downloads/Certification_FAQ_final.pdf
http://www.intel.com/netcomms/technologies/wimax/306013.pdf