Half of Internet traffic will be IPv6 by 2020
As we've crossed into the second half of 2017, it's time to take stock of how the Internet continues to evolve from the pure IPv4 network from nearly a decade ago, to the increasingly mixed IPv4-IPv6 Internet of today. Over the last few years, I've posted a summary of google's IPv6 statistics measurements and extrapolated these forward a year or two based on exponential and quadratic regression curves. Google's data represents one perspective though probably a good indicator for the Internet at large. In this post, we'll take a walk down memory lane first to see how prior years' predictions have panned out.
In February, 2014, I posted my first set of curve fitting projections as related in the table below along with actual measurements:
Clearly, the exponential curve fit was overly ambitious while the quadratic fit under-predicted the curve. Another post from July, 2015 illustrated a revised view of the prior predictions.
Despite being dampened from the prior prediction, the exponential curve fit proved to be overly ambitious. The quadratic curve again lagged actuals. From hence, I've dismissed the exponential curve as unrealistic and applied a third order polynomial curve fit to serve as a closer upper bound. The following chart illustrates the regression curve based on data up through the end of June.
Based on these predictors, IPv6 penetration, as defined here as mapping to google.com IPv6 density, could reach 25% in about one year, and 30% by the end of 2018. These projections also indicate penetration would hit 50% by the second half of 2020. In about three years the Internet could be an equally split IPv4-IPv6 Internet.
Of course, these predictions are based solely on past performance. As momentum grows, major mobile and Internet providers may take initiatives to accelerate or otherwise alter the pace of IPv6 penetration. If your universe of Internet users to whom you invite access to your website, email servers, or other Internet applications mirrors that of Google, i.e., the whole Internet, you need to begin deploying IPv6 if you have not already done so. If half of the Internet's users cannot efficiently access your website in two years, consider the lost opportunity in terms of commerce, communications, or information sharing. In reality, dual stack devices and those behind carrier grade NATs should still be able to access your site given the Internet standard approach of attempting IPv6 connections first, following by IPv4. But at some point it's likely that devices will no longer have access to an IPv4 address and will be only able to communicate via IPv6. Major ISPs have pervasively deployed IPv6 already and Android and ios mobile platforms require support of IPv6-only networks.
If you have not yet deployed or even considered IPv6 implementation, I invite you to access our free online tools to help you familiarize yourself with business drivers for IPv6, your return on investment (ROI) for deploying IPv6, as well as IPv6 addressing and subnetting. These tools can help you understand the implications both of deploying IPv6 and of not deploying in terms of upside opportunity versus cost.
In February, 2014, I posted my first set of curve fitting projections as related in the table below along with actual measurements:
End of year
|
Exponential
|
Quadratic
|
Actual
|
2014
|
5.9%
|
4.9%
|
5.7%
|
2015
|
13.8%
|
8.2%
|
10.0%
|
2016
|
32.3%
|
12.5%
|
16.8%
|
Clearly, the exponential curve fit was overly ambitious while the quadratic fit under-predicted the curve. Another post from July, 2015 illustrated a revised view of the prior predictions.
End of year
|
Exponential
|
Quadratic
|
Actual
|
2015
|
11.4%
|
8.9%
|
10.0%
|
2016
|
25.0%
|
15.0%
|
16.8%
|
Despite being dampened from the prior prediction, the exponential curve fit proved to be overly ambitious. The quadratic curve again lagged actuals. From hence, I've dismissed the exponential curve as unrealistic and applied a third order polynomial curve fit to serve as a closer upper bound. The following chart illustrates the regression curve based on data up through the end of June.
Based on these predictors, IPv6 penetration, as defined here as mapping to google.com IPv6 density, could reach 25% in about one year, and 30% by the end of 2018. These projections also indicate penetration would hit 50% by the second half of 2020. In about three years the Internet could be an equally split IPv4-IPv6 Internet.
Of course, these predictions are based solely on past performance. As momentum grows, major mobile and Internet providers may take initiatives to accelerate or otherwise alter the pace of IPv6 penetration. If your universe of Internet users to whom you invite access to your website, email servers, or other Internet applications mirrors that of Google, i.e., the whole Internet, you need to begin deploying IPv6 if you have not already done so. If half of the Internet's users cannot efficiently access your website in two years, consider the lost opportunity in terms of commerce, communications, or information sharing. In reality, dual stack devices and those behind carrier grade NATs should still be able to access your site given the Internet standard approach of attempting IPv6 connections first, following by IPv4. But at some point it's likely that devices will no longer have access to an IPv4 address and will be only able to communicate via IPv6. Major ISPs have pervasively deployed IPv6 already and Android and ios mobile platforms require support of IPv6-only networks.
If you have not yet deployed or even considered IPv6 implementation, I invite you to access our free online tools to help you familiarize yourself with business drivers for IPv6, your return on investment (ROI) for deploying IPv6, as well as IPv6 addressing and subnetting. These tools can help you understand the implications both of deploying IPv6 and of not deploying in terms of upside opportunity versus cost.
Comments
Post a Comment