1. Transition Mechanisms for Ipv6 Hosts and Routers RFC2893 By Michael Pfeiffer

2. Terminology  IP v4 only node  IP v6 only node  IP v4/IP v6 node  IP v4 compatible IP v6 Address High order 96 bits prefix 0:0:0:0:0:0 High order 96 bits prefix 0:0:0:0:0:0 Low order 32 bits are the IP v4 address Low order 32 bits are the IP v4 address

3. Techniques Used in the Transition  IP v6-over-IP v4 tunneling  Configured tunneling  Automatic tunneling  IP v4 multicast tunneling

4. Dual IP Layer Operation  Modes of Operations for IP v4/v6 nodes Both stack enabled Both stack enabled Disabling one or the other stacks Disabling one or the other stacks  Tunneling techniques None None Configured tunneling only Configured tunneling only Both configured and automatic tunneling Both configured and automatic tunneling

5. Address configuration  IP v4/v6 node an have two addresses IP v4 address IP v4 address 32 bit address32 bit address Acquired using IP v4 addressing mechanismsAcquired using IP v4 addressing mechanisms IP v6 address IP v6 address 128 bit address128 bit address Acquired using IP v6 addressing mechanismsAcquired using IP v6 addressing mechanisms

6. Domain Name System  Ability to map between both IP versions Example: What happen when IP v6 node requests a record that as been defined with a IP v4 address Example: What happen when IP v6 node requests a record that as been defined with a IP v4 address  The resolver libraries must be capable of handling both IP v6/IP v4 records

7. DNS  Returning records to IP v4/v6 nodes Filtered Filtered Return only IP v6 addressReturn only IP v6 address Return only IP v4 addressReturn only IP v4 address Ordered Ordered Return both addressesReturn both addresses May change the order of addresses two influence the receiving node. May change the order of addresses two influence the receiving node.

8. Addressing Addresses in the DNS  Before a IP v4/v6 record is added all of the following should be true: The address is assigned to the interface on the node The address is assigned to the interface on the node The address is configured on the interface. The address is configured on the interface. The interface is on a link which is connected to the IP v6 infastructure. The interface is on a link which is connected to the IP v6 infastructure.

9. Common Tunneling Mechanisms  Will take a while to build up IP v6 infastructure so we need a means of sending packets  Tunneling is a way to use IP v4 infrastructure to send IP v6 packets  This is done by sticking a IP v6 packet into the body of a IP v4 packet.

10. Types of tunneling  Configured Tunneling Router to Router Router to Router Host to Router Host to Router  Automatic Tunneling Host to Host Router to Host

11. Tunneling Configuration  1 st The encapsulating node creates an encapsulating IP v4 header and transmits  2 nd The decapsulating node receives the encapsulated packet, reassmebles the packet (if needed), and removes the IP v4 header.  Note:(The primary diffence in tunneling scheme is how the determine the header packet end address.)

12. Encapsulation +-------------+ +-------------+ | IP v4 | | Header | +------------+ +-------------+ | IPv6 || IP v6 | | Header || Header | +------------++------------+ | Transport | ====>| Transport | | Layer | | Layer | | Header || Header | +------------++------------+ | || | | Data | | Data | | || | +------------++------------+

13. Handling Fragments  What happens when a IP v6 packet is too big for the payload of IP v4 packet? IP packet is fragmented and two or more packets are send? IP packet is fragmented and two or more packets are send? This can be avoided in when the encapsulating node knows about the tunneling. This can be avoided in when the encapsulating node knows about the tunneling.

14. Fragmentation Problems  Fragmentation in tunneling not sent by the sending node  Resassembling packets at the router.

15. Reducing Fragmentation  Making packet small enough for additional IP v4 header  Using a IP v4 Path Discovery Protocal. If you have a large number of tunnels then i might not be able to store all of them. If you have a large number of tunnels then i might not be able to store all of them.  This will not completely elminate fragmentation but will reduce it.

16. Hop Limit  IP v6-over-IP v4 tunnels are considered as one hop. This means the IP v6 hop limit is decremented by one at the end of each tunnel This means the IP v6 hop limit is decremented by one at the end of each tunnel  This hide the existance of tunneling. IP v6 header only gets used at the beginning and end of the tunnel. IP v6 header only gets used at the beginning and end of the tunnel.

17. ICMP Errors  Errors are sent to the encapsulating node  ICMP “packet too big” error is handled according to the path dicovery Which will change the packet along the path Which will change the packet along the path  Handling of other errors depend of the size of the “packet in error” field sent back.

18. IP v4 Header for tunneling  Version field: 4  IP header length in 32 bit words: 5  Type of service: 0 Note (work is underway to redefine the service byte, so this might be different in the future.) Note (work is underway to redefine the service byte, so this might be different in the future.)  Total Length: (Payload + IP v6 header + IP v4 header)

19. IP v4 Header Cont.  Identification: Generated uniquely as for any IP v4 packet transmitted by the system.  Flags: Don't Fragment flag Don't Fragment flag More Fragmentsd flag More Fragmentsd flag  Fragment offset: set if there is a fragmentation  Time to Live: Set in implementation specific manner

20. IP v4 Header Cont.  Protocal: 41 (Assigned payload type number for IP v6)  Header checksum: calculated checksum for the IP v4 header  Source address: IP v4 address of encapsulating node  Destination address: IP v4 address at end of tunnel.

21. Decapsulation  Accurs when a node recieves a packet with it's IP v4 address and the protocal field is set to 41  Reassembles packet if it was fragmented  IP v6 header will be untouched since it had been encapsulated  Hop limit will be decremented by one

22. Decapsulation  Decapsulating node performs IP v4 reassembly before decapsulating the IP v6 packet Done to preserve all IP v6 options even if fragmented Done to preserve all IP v6 options even if fragmented  Node must not forward on decapsulated packet unless explicitly configured too.

23. Discarding packets  The IP v4 header should be discarded  Node should silently discard packet with an invalid IP v4 source address (eg. Multicast, broadcast)  The node should also silently discard packets with invalid IP v6 source address

24. Link-Layer Addresses  Both configured and automatic tunneling must have link-local addresses so routing protocal can operate  The Interface Identifier should be the 32 bit address that is in the IP v4 header.  The IP v6 Link-Local address for IP v4 virtual interface is FE80::IP v4 address

25. Neighor Discovery  Automatic and unidirectional tunnels are condsider undictional Neighor discovery is only used for formation of link local addresses Neighor discovery is only used for formation of link local addresses  Bidirectional tunnels use Neighbor Unreachability Detection (NUD) Packets to setup a tunnel.

26. Configured Tunneling  Determination of which tunnel to use is done by routing table Uses prefix mask and match technique Uses prefix mask and match technique  The Default Route

27. Configured Tunneling with IP v4 Anycast Address  Uses the Anycast Address to forward the IP v6 packet on Receiving node treats the address as if it is it's own Receiving node treats the address as if it is it's own After decapsulated transmits it toward the correct address After decapsulated transmits it toward the correct address Could have problem with fragmentation Could have problem with fragmentation

28. Automatic Tunneling  Allows IP v6/v4 nodes to communicate without pre-configured tunnels  Nodes using automatic tunneling are assigned IP v4 compatible addresses.  IP v4 address is globally unique as long as address is not for a private network.

29. Address Configuration  Will serve as both IP v4 and v6 addresses  Will acquire address through IP v4 address configuration protocols and then Map it to IP v6 address DHCP, BOOTP, RARP, Manual, etc. DHCP, BOOTP, RARP, Manual, etc.

30. Automatic Tunneling Operations  Only used when endpoint address is an IP v4 compatible address  A special static routing table can be setup for automatic tunneling  Must not be sent to IP v4 broadcast or multicast destinations

31. Source Address cofiguration  Which format: Native IP v6 Native IP v6 IP v4 compatible IP v4 compatible  Determines the type of traffic that will be returned.

32. Work sited  RFC 2893, http://www.faqs.org/rfc/rfc2893.html