IPv6 Network - DHCPv6
DHCPv6 (RFC 3315) is the IPv6 version of the Dynamic Host Configuration Protocol (DHCP). Because IPv6 has stateless address autoconfiguration, DHCP occupies a very different part of the landscape in IPv6, compared to IPv4. Although the details are different in the by-now-expected places (address length, use of multicasts, some streamlining), the DHCPv6 protocol itself is quite similar to the IPv4 version of DHCP. The more important differences are the way in which the protocol is used. DHCPv6 has three purposes:

A DHCPv6 client interested in an address and/or other configuration information sends out a solicit message indicating its needs to the link-local scope multicast address ff02::1:2, port 547. (Server-to-client messages are addressed to port 546.) DHCPv6 servers that receive the solicit message either directly or forwarded by a relay and are able to accommodate the request respond with an advertise message.

\The client considers the offers in the various advertise messages and directs a request message to the server of its choice. The server then replies with a reply message, confirming the address and/or configuration information. Alternatively, if the client only wants to receive configuration information and no addresses or prefixes, it can send a request-information message, and the server immediately sends back a reply message, so only half the messages are exchanged and the whole process completes much faster. The client may also us the “rapid commit” option to indicate that it wants to use the expedited procedure for address or prefix assignment if it’s pretty sure that it will take up the offer from the first DHCPv6 server that responds.

As expected, IPv6 addresses assigned with DHCPv6 come with a preferred and a valid lifetime. Sometime before this timer expires, the client sends a renew message, asking the server if it can continue to use the address.

When it has no more use for the address, the client sends a release message. There are some other messages for less common situations. To allow servers to recognize clients, each device that implements DHCPv6 has DHCP Unique Identifier (DUID). In IPv4, DHCP clients use a MAC address or user-supplied string as a Client Identifier. In DHCPv6 this is always the DUID. Devices may create their DUID in various ways, as long as the DUID is unique and not subject to change, if at all possible.

Cisco routers create their DUID based on the lowest MAC address in the system. Because even modular Cisco routers have stable MAC addresses, this works well. For hosts with removable Ethernet interfaces, the DUID should be created based on a MAC address and the DUID creation date. After all, an Ethernet card can reside in only one host at a time. The resulting DUID should be stored for further use, even after the Ethernet card in question is removed from the system.

DHCPv6 supports an authentication mechanism that allows clients and servers to interact in a secure way, so third parties can’t inject false DHCP messages or modify legitimate ones. However, this mechanism must be pre-configured manually on all servers and clients, partially negating the advantages of DHCP over manual configuration.

Comment an article   Name   E-mail    Title Available characters: 4000  Notify me of follow-up comments Enter what you see:

No comment posted