Dataset Comparison: IPv4 vs IPv6 traffic seen at the DNS Root Servers

As economic pressure imposed by IPv4 address exhaustion has grown, we seek methods to track deployment of IPv6, IPv4's designated successor. We examine per-country allocation and deployment rates through the lens of the annual "Day in the Life of the Internet" (DITL) snapshots collected at the DNS roots by the DNS Operations, Analysis, and Research Center (DNS-OARC) from 2009 to 2014.

Sources of IPv4 and IPv6 traffic to the DNS Root Servers

Figure 1 shows the number of countries sending traffic to monitored DNS root servers (dark color bars) compared to the number of countries that have received allocations of IP address blocks for both IPv4 and IPv6 (light color bars). As of 2014, 194 countries have an IPv4 allocation, and for 193 of them at least one IP address from their allocation has been observed in the DITL samples of packets reaching some DNS root servers. In contrast, of 184 countries with IPv6 allocations, addresses from only 113 (61%) have been sources of packets reaching the roots.

For several years, members of the DNS Operations, Analysis, and Research Center (DNS-OARC) have gathered traffic traces at anycast DNS roots servers they control as part of the Day In The Life (DITL) project. (From 2006-2009, CAIDA helped launch this project.) This data provides a unique view of the uptake rate of the IPv6 protocol worldwide. Our analysis presented here examines the DITL data sets collected annually between 2009 and 2014.

The Internet Protocol (IP) specifies the format and rules for communication between computers attached to the global Internet. The two deployed versions of this protocol are IPv4 and IPv6. IPv4 was initially used for the deployment of the public Internet. IPv6 was designed to address many of the perceived limitations of IPv4, the most pressing of which is the depletion of the pool of remaining unallocated IPv4 addresses.

The Internet Assigned Numbers Authority (IANA) manages IP address space globally, and five Regional Internet Registries (RIRs) manage it at the regional level. IANA allocates IPv4 addresses to the RIRs, and the RIRs in turn allocate them directly to organizations or to country level registries called Local Internet Registries (LIRs). Although the IPv4 protocol uses 32-bit addresses, implementation specifics and allocation policies reduce the number of available IPv4 addresses to much lower than the theoretical maximum of 2^32 addresses. IANA allocated the last block of available IPv4 addresses to the RIRs in February 2011 (although some IPv4 addresses have since been returned and are being reallocated). Since then the RIRs have also either exhausted their available pools of IPv4 addresses or severely restricted allocation rates. The protocol development community intended that the 128-bit addresses designed into the IPv6 protocol would eliminate the problem once IPv6 was deployed. In reality, this transition has not substantively occurred.

One way to measure the rate of IPv6 deployment is to analyze the number of countries that have at least one IPv6 address block allocation. We examined the RIR's delegation files that contain monthly snapshots of all Autonomous System numbers (ASNs) and IP addresses delegated to each country, and counted the number of countries that have at least one allocation in each file. In Figure 1, the light green bars represent the number of countries that have an allocation of at least one IPv4 block while the light blue bars represent the number of countries with at least one IPv6 allocation. From 2009 to 2014 the number of countries with IPv4 allocations grew from 187 to 194 while the number of countries with IPv6 allocations grew from 125 to 184.

Although the number of countries with IPv6 allocations is catching up (it is now 95% of those with IPv4 allocations while it was only 67% in 2009), there is still a large disparity in the rates at which these addresses are observably used. We examine this disparity using the annual DITL data sets collected at participating DNS root servers.

The Domain Name System (DNS) is a distributed hierarchical naming system used by Internet users to map between IP addresses and human-friendly hostnames. At the top of this hierarchy is a set of computers known as the DNS root servers, or "roots". Most DNS root name server operators are members of OARC, an umbrella group that brings together operators, implementors, and researchers. Some operators participate in the DITL data collection events, where on a certain date they collect traffic at the DNS roots they maintain. Many Internet transactions initiated begin with a DNS query, these collections can shed some light on the rates at which IPv4 and IPv6 addresses get used.

Figure 1 also shows the number of countries where we observe at least one IP address from those allocated to a given country at one or more instances of the root servers that participated in the OARC DITL data collections. The dark green and dark blue bars represent the visibility of IPv4 and IPv6 addresses, respectively. The fraction of countries where at least one IPv4 address from their allocation was observed sending traffic to the DNS roots relative to the total number of countries with IPv4 address allocations rose 97% in 2009 to more than 99% in 2014 (when only one country was not observed). Over the same period, the analogous fraction for IPv6 addresses grew from 50% to 61%. Although the number of countries with IPv6 deployments continues to grow rapidly, there still exists a large gap between the number of countries that have IPv6 addresses and those that appear to actually use IPv6. (A caveat: IPv4 addresses can be used to make IPv6 queries, and vice-versa, which we do not account for in this analysis.)

The remaining graphs in this report examine relevant Internet statistics at the country level: each dot represents a single country and its color shows its continent. One can interactively change the plots presented in each Figure. The description following each Figure corresponds to the default mode of switching the graphs by clicking on the arrows immediately below each plot. However, by checking the manual control box, one can manually set all parameters of the graph: the year of the data displayed^*, the variable shown on each axis, linear or logarithmic^** scale, and what the dot radius represents. Hovering above each dot reveals the presented data in detail: country name and continent, x and y values, and the value defining the dot's size.
* - if there is no data for a given year, we substitute the previous year's value instead.
** - In the logarithmic plots, zero values are drawn below the -axis line.

Figure 2 plots the number of IPv4 and IPv6 blocks seen at the DNS roots vs. those allocated to individual countries. The fraction of address blocks seen at the roots out of the total allocated addresses is 58% ± 17% for IPv4 compared to 16% ± 24% for IPv6. The much narrower scattering of dots in the IPv4 plot likely reflects the much more uniform usage of IPv4 addresses across countries.

Figure 3 elucidates how per-country IPv4 and IPv6 allocations have changed from 2009 to 2014. Consistent with Figure 1, this Figure shows that most countries had no IPv6 allocations in 2009, but most have at least one IPv6 block in 2014. The only country whose number of IPv6 allocations decreased is Monaco: it lost its first allocated block to Switzerland in 2010 and its last allocation to Serbia in 2011. Brazil went from being a slight outlier in terms of the ratio of IPv4 to IPv6 allocations in 2009 (89:1), to a more common position in 2014 (7:3), mostly because of a significant increase in the number of its IPv6 allocations.

Figure 4 examines the relationship between the number of IP addresses and the number of packets seen at the monitored DNS roots, per country. The dot's size reflects the packets/addresses ratio for a given country. Similar to the pattern observed when comparing the number of blocks observed to the number of blocks allocated, the IPv4 packets/addresses ratios for individual countries is less scattered than the IPv6 ratios. The U.S. sends more packets from more addresses then any other country in both IPv4 and IPv6. In the IPv4 space, China is the next largest user: although its number of addresses seen at the roots is only 40% of the U.S. number, its number of packets is 80% of the number of the U.S. packets.

Figure 5 presents the relationship between each country's usage of IPv4 and IPv6 address allocations and its Gross Domestic Product (GDP), which is a measure of the wealth of a country. (The GDP data are from the World Bank indicators). The dot size shows the total number of address blocks allocated to each country and tends to increase with GDP, indicating a strong positive correlation: wealthier countries have more allocated address blocks in both IPv4 and IPv6 space. However, the fraction of allocated addressed that were actually observed at the root servers does not positively correlate with GDP; in IPv4 it is negatively correlated with GDP, while in IPv6 there is no correlation. The negative correlation for IPv4 address usage vs. GDP is consistent with richer countries having a surplus of IPv4 space relative to poorer countries, and thus not feeling pressure to utilize IPv4 addresses as efficiently.

Additional Content

/archive/policy/dns-country/test/

Dataset Comparison