The Rural Broadband Initiative (RBI) funded by Government is having an immense impact, halving download times for upgraded connections, but still not as good as urban broadband.
(TrueNet is contracted to the Commerce Commission to monitor broadband performance. It uses a network of probes attached to 400 volunteers' internet connections to rate ISPs — Editor.)
The Rural Broadband Initiative (RBI) funded by Government is having an immense impact, halving download times for upgraded connections, but still not as good as urban broadband.
Older rural ADSL connections take more than twice as long to load webpages than RBI upgraded connections, and approaching 3 times that of Urban services.
In February 2015, Orcon is the standout ISP with top performances for both Fibre, and ADSL services.
Orcon Fibre100Mb/s now leads with the fastest time to download NZ Live Webpages, overtaking Snap for the first time since November 2014. Snap leads with downloading Australian Live Webpages.
Significantly, there is now very little difference in performance between all the Fibre 100 & 30Mb/s services for Live New Zealand located webpage downloads. This is not yet true of Australian webpages, where some ISPs take 50% longer than others.
For VDSL, Vodafone remains dominant for NZ Live Webpage downloads followed by Orcon, but for Australian webpages Vodafone, BigPipe and Orcon are equal best.
For ADSL services, Orcon takes the top spot, overtaking Slingshot for time to download NZ Live Webpages. This is a significant achievement as Slingshot has dominated the ADSL performance statistics for this particular measurement since November 2014.
TrueNet changed focus in February 2014 from Speed performance to Webpage performance. In twelve months the performance range between ISPs has improved as has the actual performance. Most broadband connections in New Zealand use ADSL services so it’s important to note that the ADSL performance parameters for time to download NZ Live Webpages have narrowed considerably since November 2014, when the range from best to worst was 4.7 to 6.2 seconds. In February 2015 the range is now 3.7 to 4.7 seconds. These measurements show a big improvement in download times as well as a narrowing between best and worst results.
Speed by Time of day consistency has proved to be more volatile, although ADSL, VDSL & 50Mb/s Cable have consistent 9PM download speeds.
There were changes in high speed Cable and Fibre averages this month, where Cable evening performance slowed from 109Mb/s in January, to 89Mb/s in February. Vodafone 100Mb/s Fibre service is slowly improving with November 9pm results of 35Mb/s, and February results now 49Mb/s.
Our Quarterly feature is Rural Broadband this month. No surprises that the rural sector experiences a lower performance compared with the urban sector . Rural ADSL averages less that half the speed of urban users at 9pm. Rural connections take over 2.5 times longer to download the Live Webpages in our tests.
As should be expected, RBI funded ADSL connections fare better at downloading Live Webpages, and experience less impact on performance during the 9pm busy period than yet to be upgraded Rural ADSL.
Table 1: Summary of Performance Measures
(Click to zoom)
Responsive website browsing is valued by most Internet users, and conversely, slow-loading sites can be extremely frustrating.
TrueNet tests internet browsing by downloading a selection of Live NZ and Australian Webpages, measuring the the time to fully download all files on the page. These pages are changed from time-to-time.
All Fibre services loaded the entire sample of Webpages in just over 1 second, with Orcon 100Mb/s Fibre the best this month. The relative performance of 100Mb/s and 30 Mb/s Fibre services are similar, with negligible difference.
For ADSL services, Orcon and Bigpipe are best this month, although these pages take 4 seconds to download.
New Zealand ISPs are in a different league to iiNet (Australia) which has a remarkably slow performance of 7.9 seconds to download NZ Live Webpages. Surprisingly it also lags behind NZ providers for Australian Live Webpage download times.
Flip, which used to lead this group in the past, now seems to have settled into the middle of the pack.
Chart 1: Live NZ Webpage Download Time - in seconds
Snap Fibre comfortably led the 100Mb/s services, and was closely followed by Orcon Fibre in the 30Mb/s services.
Vodafone Cable has a similar performance to Fibre in these downloads (see Vodafone in Chart 3, International Download Speed).
Of the VDSL providers, Orcon, Bigpipe and Vodafone are equal, with Spark the slowest.
With the ADSL providers Flip moved to the top spot this month, while Slingshot dropped from first to third place. Sparkimproved compared to last month, leaving iiNet (Australia) the slowest to download the 4 Australian Webpages.
For TrueNet's speed tests each panelist's probe regularly downloads a 1MB file from Auckland, Wellington, Dallas and Sydney.
For national tests, the best Auckland or Wellington download speed result is used from each test run to compute a median hourly speed for each panelist's connection. International tests represent the median hourly speed of each probe from Dallas and Sydney.
TrueNet tests download speeds from our servers in Dallas & Sydney to measure performance to our international neighbours.
We ensure the download file is not held in New Zealand (cached), so that the test truly measures international performance. The speed test results selected are taken at 9pm, in the evening busy period, when congestion is likely to be worst.
Orcon 100Mb/s Fibre is the fastest to Sydney, approaching 70Mb/s. By comparison, the NZ national speed at 9PM was 80Mb/s this month, thus users are achieving excellent relative performance to Sydney.
Vodafone consistently has the fastest speed to Dallas with their 100Mb/s Fibre service, but note that the legacy Cable services do not share this performance.
Vodafone continues to have the best ADSL/VDSL speeds as a percentage of the NZ Speed from both Dallas and Sydney while Slingshot is a close second. The gap between ISPs is evident every way we review this data, including comparing actual speed. The difference between ISP average connection speeds is minimal in either ADSL (~10Mb/s) or VDSL (~22Mb/s).
Chart 4: ADSL, VDSL File Download Speed - Dallas & Sydney
Comparing performance by time-of-day is important as it shows the service degradation when everyone is using the internet during the peak hours of 8pm to 10pm. TrueNet uses the best of a pair of Auckland and Wellington download tests to calculate the median results by hour over the month for each monitored connection. We take the average of all results with each ISP for each hour.
Download speed by Time of Day is useful to reveal service degredation during periods of high use, typically between 8pm and 10pm.
Vodafone's 100Mb/s Cable service still leads all other services although with a pronounced evening period dip in median speed, for the first time since the service offer was reduced from 130Mb/s the average speed at 9pm dropped below 100Mb/s.
Snap and Orcon Fibre 100Mb/s services are running below their 100Mb/s advertised speed, with Orcon showing a little more variability through the 8pm busy time.
Vodafone Fibre 100Mb/s service is consistently running well below their advertised speed, and shows very little difference in speed to their 50Mb/s service. We have one panelist that runs at almost 100Mb/s, the only one not connected via Chorus.
All Fibre 30Mb/s services are stable, and are running true to their advertised speed.
Chart 5: Fibre, Cable, and VDSL Download Speed
TrueNet has fibre panelists on 8 different service speeds. An alternative way to report results is to compare performance against advertised service speed.
In Chart 6, the performance of ISPs with 5 or more probes is shown. Spark leads this month, closely followed by Snap. Fibre 30Mb/s services all perform very well independent of ISP, but we have only 2 Vodafone panelists on this speed which means Vodafone is dominated by 100Mb/s fibre in Chart 6. TrueNet has more 30Mb/s than 100Mb/s panelists with Spark, and Orcon but their other speeds performance is also closer to advertised. We have amost equal numbers of Snap 30Mb/s and 100Mb/s panelists, with some other speeds such as 50Mb/s and 1000Mb/s.
Chart 6: Fibre Performance as Percentage of Advertised Speed
VDSL File Download Speed
There is more variation in the Time of Day measurements this month, though most results are still very good with Orcon and Bigpipe showing as top performers.
The evening period shows more contention than last month, with Slingshot dropping well below the target performance of 95% by reaching just 84% at 9pm, we expect that this speed reduction is very noticeable. Vodafone and Snap also drop below our target of 95% at 9pm.
Snap are clearly having speed issues in the wee small hours, probably due to there "All you can eat nights" plan between 1am and 7am. We do not identify users on this plan and Snap assure customers in their "offer summary" that they do not traffic manage, so this measures the impact on all customers, however at 95%, speeds remain within our target.
Chart 7: VDSL File Download Speed by Time of Day
ADSL File Download Speed
Overall an excellent result for ADSL test, with the exception of iiNet and Slingshot. The impact of growth of demand on supply is demonstrated by Flip which has now dropped below 95% at 9pm, having been top performer in November 2014 with 99%.
Orcon achieved the best result in ADSL as well as VDSL, closely followed by Bigpipe and Spark.
Chart 8: ADSL File Download Speed by Time of Day
Upload speed is important when sending files and content to the internet. Upload speed impacts the time taken to upload a photo or synchronise data with the cloud.
TrueNet's upload test sends a 1MB file to our server in Wellington, and records the results using a similar method to the download tests, but measured from our Wellington server only. Australian connections are tested to our Sydney server.
The aggregate speed of 50Mb/s Services remained at 44Mb/s this month.
The 20 Mb/s Fibre service averaged 18 Mb/s.
Cable and Fibre 10Mb/s consistently average 10Mb/s, or close to it.
VDSL averaged 8.5Mb/s, significantly lower than Fibre or Cable, but still much better than ADSL.
ADSL crawls along at just 1Mb/s, almost too slow for video calls at the far end.
Unlike download tests, Time of Day variations are minimal in upload testing, implying that there is little contention in the access, or backhaul networks.
Chart 9: Upload Performance by Technology
Snap continues a trend from last month delivering the top speed, above its advertised 50Mb/s.
The under achievers for February are Orcon 50Mb/s Fibre, and Snap 20Mb/s Fibre, neither reaching their advertised speed.
Chart 10: Fibre, Cable Upload Speed by ISP
With the government announcing expansions to the Rural Broadband Initiative (RBI), it is timely to review the results from our rural probes. Rural areas are measured with the same methodology as the Urban results, using the best of Auckland & Wellington 1MB file downloads for speed tests, and the same set of Australian and NZ Live Webpages.
The RBI funded ADSL panelists again experienced superior speed and Live Webpage results, which are not far off the average of Urban ADSL. Hence, RBI investment does provide significantly improved services in rural areas.
Webpage Downloads
The Live Webpage test results are shown in Chart 11. Rural Funded RBI, although slower to load than the average Urban ADSL, is not far different. Rural ADSL, possibly due to copper lengths, averages more than twice as long to load webpages compared to RBI Funded services, and approaching 3 times that of Urban services.
Chart 11: Rural vs Urban - Live Webpage Comparison
File Download Performance of Rural
Due to migration of volunteers, only Vodafone and Spark ADSL had sufficient panelists to report on separately, both had similar results. RBI Funded ADSL are shown as a single group, which includes panelists from Slingshot, Spark, Vodafone, and Xnet.
File download test results for Rural ADSL users in Chart 12, show large speed reductions during the evening, significantly more so than for Urban users. The implication is that, lack of speed is an issue for Rural ADSL, as is contention.
Spark and Vodafone fare somewhat worse than the composite of remaining rural ADSL panelists (Rural ADSL Other), however all rural connections report well below our 95% target demonstrating that a lot of the congestion is beyond the control of the ISPs.
RBI Funded ADSL is also impacted much worse than our target during the evening period, the results are much better than rural ADSL but would show well below the majority of ISPs in our Urban comparison in Chart 8.
