IPv6: Addressing the Future

1. IPv6:Addressing the Future Fred Baker Cisco Fellow

2. Points to ponder • The past: where networks came from • The future: where networks are going • IPv6 innovations: what is really different? • IPv6 debate: is IPv6 really a sufficient solution? • IPv6 today: status in implementation and deployment

3. The Past

4. Diverse networks joined by application-layer gateways Inevitable loss of functionality crossing proprietary application and network boundaries Difficult to deploy multi-network applications Hard to diagnose and remedy problems Stateful gateways inhibited dynamic routing around failures No global addressability Ad-hoc, application-specific solutions Before IP

5. Simple Applications, Smart Network Able to provide high quality services to specific applications Network does one thing well: deliver specified services to specified applications Intolerant of change Simple Network, Smart Applications End to End Principle Network does one thing well: ship packets Applications can do anything that can use that paradigm Fundamental Premises:

6. IP routers & global addresses Simple, application-independent, least-common-denominator network service: best-effort datagrams Stateless gateways could easily route around failures With application-specific knowledge out of the gateways: Anyone could deploy new, internet-wide applications and services Internet became a platform for rapid, competitive innovation The IP Solution

7. Network address translators and application-layer gateways Inevitable loss of some functions Difficult to deploy new internet-wide applications Hard to diagnose and remedy problems Stateful gateways inhibit dynamic routing around failures No global addressability Ad-hoc, application-specific (or ignorant!) Solutions The Internet Today

8. The Future

9. The Probable Future • Billions and billions of new Internet devices • Billions of new Internet users • Internet available everywhere, all the time (wired, wireless, mobile,…) • Convergence of all communication on the Internet (business, personal, entertainment, public services,…)

10. Continued degradation of the end to end model with IPv4? More complex and volatile network service => Lower performance, less robust, less secure, less manageable More centralized control over new applications and services => Significant barrier to innovation and growth The Unknown Future

11. …or restoration of the end to end model with IPv6? Simple, stable network service => Higher performance, more robust, more secure, more manageable Enabling anyone to provide new applications and services => Allowing rapid innovation and growth The Unknown Future

12. IPv6 Innovations

13. Lots of Addresses • IPv4 Internet: O(232) = 4,294,967,296 addresses • Arbitrary division into networks • 12.5% allocated to non-host addresses • ~45% allocated to various networks • ~26% advertised in today’s Internet • Conservatively allocated • IPv6 Internet: O(2128) = 3.4*1038 addresses • O(264) = 18,446,744,073,709,551,616 Networks • O(264) = 18,446,744,073,709,551,616 hosts per network • Host addresses self-allocated Enough!

14. Plug-and-play • One of the nice things about AppleTalk: • You can plug the device or computer in, and it just works • One of the not-so-nice things about IPv4: • You can plug the device or computer in… • Configuring, and reconfiguring, can be hard • DHCP helps a lot, but it requires properly configured servers • IPv6 allows for • Significant level of autoconfiguration • Automated network renumbering • Arbitrary device addressing within topological limits

15. IPv4 Mobility Permits device to move using same home address All communication through Home Agent Foreign Agent must be a router IPv6 Mobility Permits device to move using same home address Communication via care-of address No Foreign Agent required Security Issues: Session hijack Duration of Switchover Mobility Dogleg Routing Optimized Routing

16. Addressing and Naming of Applications One of the nice things about NetWare: Service Location Today: DNS lists several addresses for a name, but no information to help select a server Anycast

17. Proposal: DNS lists one address, Servers are “routers” to that address DNS for service name, Common address for service location Topological address for specific access Issues: Route changes can change which server you use in mid-transaction Solution: Treat server as a mobile device which is currently stationary Connect to “home address” to select server, Thereafter talk to fixed “care-of address” Anycast

18. Security issues • IPv6 enables end-to-end use of IPsec protocols (because it eliminates NATs), • Plus for security, although IPsec also exists in IPv4 Internet and is widely used for VPNs • Authentication (“you are the person who knows this key”) • Antidote to session hijack (“you are the same person I was just talking with”) • Privacy (encryption, using symmetric or public key technology) • IPsec authentication dependant on key distribution infrastructure, which is not currently a solved problem • Affects mobility, anycast, secure communication in general

19. IPv6 Debate Geoff Huston’s questions

20. Are we really running out of addresses? • Growth in IPv4 advertisement rate not high • But folks who need addresses can’t get them • Largely a question of perspective • If you already have your addresses assigned, getting them is not a worry

21. Everyone doesn’t want a permanent address • Everyone who wants one is not able to get one • Example: People’s Republic of China • 1.3 Billion people • Order of magnitude growth in Internet usage year over year • ~9M addresses in 1999 • ~16M addresses in 2001 • Do we simply assume that anyone who has not already asked never will? • Africa, South America, India, Arab world…

22. In client/server applications Clients vastly outnumber servers Clients can be addressed on demand Examples: WWW, FTP, X-Windows But every application is not client/server Peer/peer applications Peer must be accessible and addressed when someone decides to talk with it Do we want to limit ourselves to the client/server model? Every device is not a server

23. Privacy? • Privacy issues • Concern: inclusion of MAC address in IPv6 breaks privacy • Reality: 1:1 correlation between IP and MAC Address breaks privacy in either IPv4 or IPv6 • Privacy solutions in IPv6 • Autoconfiguration procedures enable, for example • Random address changes every hour • Address per user of multi-user machine • Address per TCP session or per web page loaded

24. Number of usable addresses • Argument: • IPv4+port gives 248 effective addresses • IPv6 allocation gives 248 networks, 216 subnets, and a few hosts in each subnetwork • Comparable when viewed on the service provider network • Not really comparable • Math error: 248¹ 248+16 • Not responsive to user network design issues

25. Is it enough better to justify changing? • Argument: • IPv6 doesn’t change routing, trust model, QoS, etc • It gives us IPv4 Internet with more addresses • What IPv6 does do: • Removes address conservation as an issue • Enables kinds of applications current addressing makes difficult • Simplifies deployment of new applications • Eliminates need to kludge around addressing issues

26. IPv6 Today

27. Standards • Core IPv6 specifications are IETF Draft Standards=> well-tested & stable • IPv6 base spec, ICMPv6, Neighbor Discovery, PMTU Discovery, IPv6-over-Ethernet, IPv6-over-PPP,... • Other important specs are further behind on the standards track, but in good shape • Mobile IPv6, header compression, A6 DNS support,... • For up-to-date status: http://playground.sun.com/ipng • UMTS R5 cellular wireless standards mandate IPv6

28. Implementations • Most IP stack vendors have an implementation at some stage of completeness • Some are shipping supported product today,e.g., Cisco, 3Com, *BSD(KAME), Epilogue, Ericsson/Telebit, IBM, Linux community, Hitachi, Nortel, Sun, Trumpet • Others have beta releases now, supported products soon,e.g., Compaq, HP, Microsoft • Others rumored to be implementing, but status unknowne.g., Apple, Bull, Juniper, Mentat, Novell, SGI • (see http://playground.sun.com/ipng for most recent status reports) • Good attendance at frequent testing events

29. Deployment • Experimental infrastructure: the 6bone • for testing and debugging IPv6 protocols and operations(see www.6bone.net) • Production infrastructure in support of education and research: the 6ren • CAIRN, Canarie, CERNET, Chunahwa Telecom, Dante, ESnet, Internet 2, IPFNET, NTT, Renater, Singren, Sprint, SURFnet,vBNS, WIDE(see www.6ren.net, www.6tap.net) • Commercial infrastructure • Some ISPs (IIJ, NTT, SURFnet, Trumpet,…) have announced commercial IPv6 service or service trials • Japan and China have announced national direction

30. Deployment (cont.) • IPv6 address allocation • 6bone procedure for test address space • Regional IP address registries (APNIC, ARIN, RIPE-NCC)for production address space • Deployment advocacy (a.k.a. marketing) • IPv6 Forum: www.ipv6forum.com

31. Conclusions • IPv6 is addressing the future… • Addresses for new devices, new applications, and new users • Restoring the end to end model, for performance, robustness, security, manageability, and enabling rapid innovation • Enhancing IP for next-generation applications: multicast, mobility, plug-and-play, security, and multiple qualities of service • …but is it a future we will see? • Must apply much more energy, in design, implementation, deployment, transition, training, explaining,… • The only way to fight entropy is to apply energy