![[CAIDA - Center for Applied Internet Data Analysis logo]](/images/caida_globe_faded.png "Go to CAIDA home page")
This page summarizes our exploration of the evolution of IPv6, supported by NSF grant CNS-1111449, "NeTS-IPv6: Exploring the evolution of IPv6: topology, performance, and traffic".
Principal Investigators
- kc claffy (PI, CAIDA)
- Robert Beverly (Co-PI, Naval Postgraduate School)
Goals of the project
The high-level goal of this project is to measure the evolution of IPv6 in three dimensions: topology, traffic, and performance. Despite significant publicity in industry related to IPv4 address run out and the long term benefits of transition to IPv6, multi-year IPv6 traffic volumes are commonly reported to be less than 0.1% of traffic. We seek to uncover characteristics of current IPv6 deployment that can be used to infer how to advance IPv6 deployment -- be it rooted in technical capability or policy development.
Our approach focuses on acquisition and multi-dimensional analysis of the following properties of the IPv6 Internet extracted from relevant data (listed below) across time.
- Structure and dynamics of the IP and AS-level topology graphs
- Deployment, usage, and performance of transition technologies
- Workload characteristics and trends: who is sourcing what types of IPv6 traffic
- Relationship of topology, geography, and network type to deployment usage and performance
- Impact of middleboxes and protocol functions (path MTU discovery, ECN, SACK) on reachability and performance
Data collection
| Data type | Description | Source |
|---|---|---|
| Interdomain traffic | IPv6 traffic workloads on a US major backbone can be used to determine the extent of global IPv6 usage. | Major US backbone |
| Interdomain topology | BGP data from Routeviews and RIPE can be used to construct AS-level Internet topology graphs | Routeviews, RIPE |
| Interdomain topology | Traceroute data from Ark can be used to infer IP- and AS-level connectivity and performance | CAIDA's Ark infrastructure |
| TCP behaviour | Special tests run on Ark systems can be used to measure the performance of TCP and infer the presence of middleboxes interfering with TCP features such as ECN and SACK, as well as IP features such as Path MTU discovery. | CAIDA's Ark infrastructure |
| AS relationships | AS relationship data derived from analysis of IPv4 BGP routes can be used to study relationship structures in the IPv6 Internet. | CAIDA's AS Rank system |
| AS classification | Classifications of ASes by their type (transit provider, enterprise customer, content/access/hosting provider) can be used to study trends of IPv6 penetration by network types. | [Dhamdhere et al 2012] |
| DNS names | DNS data from Ark can be used to infer cases where IPv4 and IPv6 refer to the same system or interface, aiding the study of relative dual-stack performance | CAIDA's Ark infrastructure |
| Edge measurements | Specially constructed javascript embedded in webpages can be used to infer IPv6 capability and performance at the edge of the network. | [Aben2011jul] |
| Edge measurements | Specially constructed honeypot infrastructure can be used to opportunistically used to infer IPv6 capability and performance at the edge of the network, and can gain insights into any growth in IPv6 capabilities of malware, scam hosting, and botnets. | [Beverly2012] |
| IP Geolocation | Freely available data from MaxMind can resolve the locations of 6to4 and Teredo IPv6 prefixes to cities, and provides country-level granularity for the reminder. | MaxMind GeoLiteCityv6-beta |
| Dual-stacked web sites | Freely available data from Alexa on the names of popular web sites can be used to automatically identify dual-stacked web sites and provide a set of targets for which we can measure and compare dual-stack performance. | Alexa Top 1M sites |
Project Deliverables
We will disseminate -- through publications at conferences, journals and network operator venues such as NANOG -- insights obtained from the measurements and data.
The schedule of work below shows categorized tasks to be completed in association with this project.
| Task | Description | |||
|---|---|---|---|---|
| Infrastructure Support | ||||
| Task 1 | Create project web pages and start regular updates | |||
| Task 2 | Evaluate IPv6 capability at Ark hosting sites | |||
| Performance of Dual-Stack and Transition Technology | ||||
| Task 3 | Enable IPFW (software firewall) capability at Ark hosting sites to enable TBIT-style measurements | |||
| Task 4 | Develop and deploy methodology to automatically test the TCP capabilities of IPv6 webservers over time | |||
| Task 5 | Use TBIT-style tests to track the evolution of middlebox impediments in the public IPv6 Internet | |||
| Task 6 | Develop and deploy tools to track the dual-stack performance of webservers over time and identify more optimal routes where possible | |||
| Task 7 | Monitor 6to4 and Teredo relay placement relative to Ark hosting sites | |||
| Task 8 | Study client tunnels and other transition technologies to identify impact on performance and adoption of IPv6 | |||
| IP-level Topology | ||||
| Task 9 | Implement system to automatically update our IPv6 probing target list | |||
| Task 10 | Implement new IPv6 topology measurement algorithms to improve accuracy and coverage | |||
| Task 11 | Research methods to infer IPv6 router aliases | |||
| Task 12 | Evaluate the applicability of IPv4 inferred AS relationships to observed and inferred IPv6 AS-level paths | |||
| Task 13 | Upgrade DNS hostname-IP database to support IPv6 | |||
| BGP-level Topology | ||||
| Task 14 | Build tools to enable tracking of the evolution of IPv6 AS core and AS rankings | |||
| Task 15 | Analyze periodic snapshots of BGP data to investigate how the structure of the IPv6 topology evolves over time by AS classification | |||
| Task 16 | Analyze BGP data to investigate economic evolution of the IPv6 Internet and contrast with the current economic evolution of the IPv4 Internet | |||
| Edge Measurement | ||||
| Task 17 | Develop and deploy JavaScript-based edge measurement at diverse third-party web sites in research, industry, enterprise, and government environments | |||
| Task 18 | Develop and deploy honeypot-based infrastructure to monitor and characterize security-related IPv6 behavior | |||
| Task 19 | Develop and deploy tools to extract IPv6 data from packet traces, including data from DNS root servers | |||
| Task 20 | Investigate the accuracy of publicly available IPv6 geolocation data | |||
| Task 21 | Correlate illuminated edge with known policy efforts to spur IPv6 deployment | |||
| Correlation with Socioeconomic Parameters | ||||
| Task 22 | Correlate IPv6 deployment by transit providers with AS type (e.g. dial-up, DSL, cable, enterprise) | |||
| Task 23 | Build tools to automate tracking of IPv6 prefixes from the time they are allocated by an RIR, correlating data with observable characteristics of prefix owner | |||
| Task 24 | Build tools to identify IPv4 prefixes that have been allocated but not observably used on the public Internet in years, and visualize changes after IPv4 exhaustion | |||
Prior Related Work by the Investigators
- Kenjiro Cho, Matthew Luckie, Bradley Huffaker, "Identifying IPv6 Network Problems in the Dual-Stack World, Proceedings of the ACM SIGCOMM workshop on Network troubleshooting (NetT '04), Aug 2004.
- Matthew Luckie and Ben Stasiewicz, "Measuring Path MTU Discovery Failures", Proceedings of the Internet Measurement Conference (IMC), Nov 2010.
- Robert Beverly and Arthur Berger and Geoffrey G. Xie, "Primitives for Active Internet Topology Mapping: Toward High-Frequency Characterization", Proceedings of the ACM SIGCOMM conference on Internet Measurement, Nov 2010.
- Amogh Dhamdhere, Constantine Dovrolis, "Twelve Years in the Evolution of the Internet Ecosystem", IEEE/ACM Transactions on Networking, vol. 19, no. 5, pp. 1420--1433, Sep 2011.