Domain Name Server (DNS), and Latency Performance
In this quarter there is a trend for some ISPs to have exceptionally large latency. This is caused by the very long path that data is sent on its journey to Sydney or Melbourne from New Zealand (via Hong Kong, or LA); and from the UK to New Zealand (via Hong Kong, then USA). At present, traceroute monitoring from our panelist's connections show the problem lies with Vodafone and 2Degrees, with Voyager caught up in similar delays due to their use of Vodafone's international connections.
Another 100% result, this time from first time reported Voyager 100Mb/s fibre panelists, who saw the 100Mb/s advertised speed or better at all times of day. There are now four broadband products where actual speed during March was faster than advertised speed at all times of day. These are:Vodafone 100Mb/s Vodafone 50Mb/s cable Voyager 100Mb/s fibre 2Degrees 30Mb/s fibre
Performance for all other products remains similar to previous months, continuing the trend of ISPs responding to having their performance tested and published continuously.
The continuing development of internet connections in the latest quarter has seen consistent improvements in performance. The Rural Broadband Initiative (RBI) is growing by about 7% per quarter, these improvements are from wireless connections over Vodafone's 4G mobile links or Chorus' fibre to rural cabinets.
This quarter, Wireless RBI performance has reduced in the evening busy hour, but with higher speeds, this may not be noticeable.
Higher speed services, 100Mb/s and above, are showing signs of congestion for panelists on smaller ISPs, evidenced by speed reductions in the evening busy hours. Larger ISPs are still performing well, and appear to have congestion under control.
NZ website download times increased across all 2Degrees technologies compared to January, despite staying the same for Australian websites.
Spark 100Mb/s panelist tests have reached a sample size large enough to be reported separately, they have until recently been include in "Other ISPs". Spark results were better than many of the smaller ISPs, so "Other ISPs" as a group without Spark panelists now show a worse result.