Metric Descriptions

1 – Fraction of ASes announcing v6 prefix
2 – v6-related queries to .nz root servers
3 – Number and type of ISP v6 offerings
4 – Service availability of top NZ domains
5 – Service availability of .nz ccTLD
6 – v6 glue record service availability
7 – confirmed v6 traffic flows from ISPs
8 – End-user v6 capability
9 – v6 traffic levels on [WIX | large ISP]

1: Fraction of ASes announcing v6 prefix

Source: RIPE – http://v6asns.ripe.net/v/6.csv?s=_ALL;s=NZ

This RIPE set of global measurements tracks the percentage of networks (ASes) that announce an IPv6 prefix for a specified list of countries or groups of countries. It provides a lead indicator of infrastructural IPv6 readiness, ie if NZ-based ASes aren’t announcing v6 addresses to their peers, there will be limited ability to exchange v6 traffic between NZ and other economies.

Note: ASN announcement of v6 addressing, does not necessarily (indeed, will usually be higher than) actual active v6 peering. Peering is not the same as traffic exchange.

This metric contributes to desired insights:

  • 1, 2 – by indicating level of v6 readiness from AS owners, including ISPs and major providers
  • 4 – by being able to easily compare to other economies

And fits to the principles as follows:

  • the data covers the whole Internet and so allows for comparative analysis
  • the website is publicly and freely available and from an authoritative source (RIPE)
  • data is presented graphically over time
  • underlying data is directly and openly available from RIPE

2: v6-related queries to .nz root servers

 Source: NZRS analysis

NZRS operates the authoritative DNS servers for .nz. NZRS are able to query their systems to determine the total number of queries of all types and whether these queries came via v4 or v6.

From this we can extract two useful (sub)metrics

  • The fraction of queries coming via v6: this gives insight into the nature of New Zealand’s ISP infrastructures and their interal level of v6 deployment
  • The volume of queries being received over v6 along with AAAA queries over v4: this gives insight into the gap between client operating system readiness and ISP connection readiness

This metric contributes to desired insights:

  • 2 – by providing insights into ISP infrastructural readiness and deployment
  • 3 – by indicating actual and relative preference of AAAA requests
  • 4 – if other economies undertook equivalent analysis

And fits to the principles as follows:

  • Historical data (for at least 18 months) will be available from the outset
  • Data generated is a result of actual queries back to the DNS root servers. Whilst this doesn’t equate to actual end-user usage, it does indicate how advanced deployments of v6 are.
  • This approach has been executed in a few other economies, and any ccTLD would be able to replicate it.

3: Number and type of ISP v6 offerings

 Source: Manual, periodic, desktop survey by TaskForce

Readiness of infrastructure metrics, like 1 and 2, are good indicators of v6 readiness, and often a leading indicator to actual v6 services being marketed and sold by ISPs (and equivalents).

This metric will take the form of a desktop survey performed on a periodic (suggested quarterly) basis and will log, for each ISP (or equivalent):

  • whether wholesale / business IPv6 services are available routinely
  • whether consumer / small business IPv6 services are available routinely
  • whether the ISP is a Large, Medium, or Small provider

Available Routinely in this case means that information regarding IPv6 (addressing, configuration, service availability) is available publicly on the the provider’s website.  Where the particular ISP is assessed as not serve a market, an entry of n/a will be recorded.

This metric contributes to desired insights:

  • 2 – by measuring marketed availability of v6 services to organisations and consumers
  • 4 – subject to additional manual analysis

And fits to the principles as follows:

  • not immediately comparable across economies unless the TaskForce undertook to analyse those markets, or similar exercises were undertaken and made available elsewhere
  • as part of the TaskForce’s work, the data would be made publicly available – and over time would provide trends
  • No historical data is available

4: Service availability of top NZ domains

 Source: Eric Vyncke – http://www.vyncke.org/ipv6status/detailed.php?country=nz

This metric leveraging the efforts of Eric Vyncke, Belgium-based Cisco employee and member of the Belgium IPv6 Taskforce equivalent (IPv6 Council). Eric has been running a site for some time that monitors the top domains in almost all countries for v6 availability.

We have worked with Eric to add an additional 500+ of the most important NZ domains to his service and categorised them by sector: ISP, Bank, Commerce etc. The list includes mostly .nz domains, but also other gTLD/ccTLD domains, eg .com, .net, .au.

The method for selecting which domains to add was as follows:

1. Get base list of URLs / organisations

a: Alexa top 150
b: lists from www.geek.nz and Mark Prior IPv6 tracking sites
c: All universities, polytecs, CRIs, government and local government sites (available from data.govt.nz)
d: All NZ registered banks (via RBNZ)

2. Manually clean and categorise using Google Refine

3. Add in any other priority organizations identified in the TaskForce’s 2011 survey of Government readiness not already present

4. Reduced  the list to c. 500 priorities through a combination of:

a. taking the primary domain only (unless a particular subdomain was deemed important)
b. pick representative sample from each category (covering largest, middle, smallest)
c. unless the total sector was small, eg, universities, banks
d. keep all ‘identified priorities’ for survey

This metric contributes to desired insights:

  • 1 – status of content providers / demand side important to New Zealand users
  • 2 – extend of ISP enablement (if we can assume that the service providers offer v6 services to their customers if their own services are v6 enabled)
  • 4 – by allowing benchmarking to other economies being measured on the same site

And fits to the principles as follows:

  • the measurements are of actual service availability (www, mail, nameservers) to the end-user
  • almost every other country is covered by Eric’s site, allowing (non automated) comparison
  • the website and data is publicly and freely available
  • analysis is supported by sector of the economy and the results for individual organisations is visible
  • some graphical historical data is maintained, but not extensive trend information

 5: Service availability of .nz ccTLD

 Source: TaskForce analysis

This metric is similar to 4, but instead of testing www, dns and mx readiness of a certain number of specified domains of interest to New Zealanders (irrespective of gTLD / ccTLD), it analyses the entire .nz space and tests for NS, MX, A and AAA records along with testing for www site presence.

The data returned would be the number of records scanned and the number that had a relevant v6 service available (for www, dns and mx).

This metric contributes to desired insights:

  • 1 – by demonstrating the availability status of content and service providers
  • 4 – by allowing comparison to other jurisdictions if the ccTLD has undertaken similar analysis, eg .fr

And fits to the principles as follows:

  • the measurements are of actual service availability to end-users
  • there are a limited amount of economies doing such analysis (we think .fr and possibly .nl and .uk)
  • historical data will be available for up to 18 months prior to the start of this benchmarking process
  • we get a picture of the entire .nz space at the average level, which provides a complimentary picture to metric 4

6: v6 glue record service availability

Source: http://dnc.org.nz//story/authorised-registrars

ISPs offering transit of v6 traffic and access to v6 DNS is an important component of service provider readiness, but the ability of domain name owners to register v6 glue records with their registrars is required to make their services visible on v6.

DNCL already surveys its registrars for this capability and presents the information on a public web-page.

This metric contributes to desired insights:

  • 1 – indicating the state of supporting market services for v6 domain advertising
  • 2 – indicating whether it is possible for content and services to be advertised on v6

And fits to the principles as follows:

  • the data is freely and publicly available on DNCL’s website
  • whilst no historic data is publicly available, it may be possible to produce some with the support of DNCL
  • the metric demonstrates actual implementation of capability with registrars, although doesn’t indicate level of uptake of services (this is addressed by metrics 2a and 2b)
  • we are not aware of any similar international disclosure at this level

7: confirmed v6 traffic flows from ISPs

Source: M-Lab – http://www.google.com/publicdata/overview?ds=z6q5492nj009fe_

Ultimately, the readiness of ISPs and hosting providers is demonstrated by proven traffic flows from end-users using their services.

Google, in partnership with other organisations, has been running M-Lab for a number of years as an “open platform for researchers to deploy Internet measurement tools”, measuring various characteristics of user connections via a number of browser-based tests.

One of the available tools (NDT) has the potential to have IPv6 turned on.  An initial NZ-based M-Lab NDT server has had IPv6 enabled (as of January 2012)

This metric contributes to desired insights:

  • 3 – by demonstrating actual deployment and uptake of v6 services by the end-user
  • 4 – if successful and deployed globally, would all comparison on all economies and on additional performance measures

And fits to the principles as follows:

  • as a new, and potentially evolving, data set there would be little historical data available
  • global deployment and hence comparability will not be available initially
  • all M-Lab results are publicly available and down-loadable via the Google Public Data explorer
  • by measuring actual end-user experiences the M-Lab tests provide the right outcome-level of v6 measurement, and have the potential to provide broader insights as the scale of usage grows.

8: End-user v6 capability

 Source: APNIC – http://labs.apnic.net/ipv6-measurement/Economies/NZ/

Earlier in 2011, APNIC Research Labs started measuring actual IPv6 availability across its membership by buying graphical advertising for web properties. This deployed a flash object in the browser that (given time) provided statistics back to APNIC. This metric does not include auto-tunnels: it’s either v6 native from ISP, or a manual tunnel. The metric is not looking at end clients capable of v6 but at the number of people who are capable, and enabled, and have v6 delivered. It is not a measure of actual traffic.

This metric contributes to desired insights:

  • 3 – by demonstrating actual deployment and uptake of v6 services by the end-user
  • 4 – as APNIC is doing this for all their economies

And fits to the principles as follows:

  • the data provides historical trends from early in 2011, in a graphical and downloadable data format
  • the website is publicly and freely available and from an authoritative source [APNIC]
  • the measurements are of actual service availability (v6 connectivity)
  • all APNIC economies are included in the measurement for comparison

9. v6 traffic levels on [WIX | large ISP]

Source: [none yet]

We are aware that both WIX and a number of ISPs are still evolving their IPv6 traffic measurement capabilities and hope to work with one or more of them to bring a measure of absolute IPv6 vs IPv4 traffic flows (as is currently done at AMS-IX and DE-CIX)

This metric would contribute to the desired insights:

  • 3 – by demonstrating actual levels of v6 traffic across a major domestic peering node
  • 4 – by allowing for comparison with other, similar, peering node measurements

Leave a Reply