DHCPv6 Dynamic Reconfiguration

Some networks are expected to support IPv4-only, dual-stack, and IPv6-only hosts at the same time. Due to devices capabilities and available connectivity types, providing generic configuration from a DHCP server to connected hosts is sub-optimal in most cases, and may even break functionality in some cases. Network infrastructure is usually well equipped to be aware of single/dual-stack nature of hosts. The network can also track and detect transitions from single to dual-stack or vice-versa. This specification describes a DHCPv6 extension for relay agents to indicate host characteristics pertaining to host connectivity to DHCPv6 servers. The information passed by a relay is generic and a DHCPv6 server can interpret and process this information to make a more informed decision on the configuration parameters that a client is to receive. The DHCPv6 server can either be configured or have built-in logic to use this information as desired, which is outside the scope of this document. describes a typical problem that can be addressed using the mechanism described in this specification. A DHCPv6 server makes a decision on priority of DNS servers to be sent back to the client based on host connectivity characteristics provided by the relay agent. While the host stack can be upgraded to send this information to the DHCPv6 server on its own, a generalized upgrade of all DHCPv6 client implementations on all operating systems is extremely difficult. [DISCUSSION NOTE: A companion solution could be to define a container that can be used to return per-AF specific configuration parameters to the client. In such a scheme, the server blindly returns all pieces of configuration and it is up to the client to make use of appropriate set of parameters according to its available connectivity. This alternative assumes an update of dhcp client. This approach can be seen as complimentary to the one defined in this specification. The document will be updated to reflect consensus of the WG on whether the additional option is to be specified.]

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in . Dual-Stack host: Denotes a host that is configured with both an IPv4 address and IPv6 prefix and is reachable using both IPv4 and IPv6 connectivity.

Default address selection rules specified in prefers IPv6 over IPv4. If a dual-stack host is configured to use a DNS64 server , it will send its DNS queries to that DNS64 server which will synthesize a AAAA response if no A record is found. Thus, the dual-stack host will always use IPv6 if a DNS lookup was involved, even if IPv4 could have been used more optimally. In some deployments, if NAT44 and NAT64 are deployed on the same network, it is preferable to use NAT44 over NAT64 because of scale, performance and application incompatibility issues (e.g., FTP) . At the same time, native IPv6 can still be preferred over IPv4. A DHCPv6 Relay Agent can observe host characteristics on a network to determine if a host is IPv4-only, dual-stack or IPv6-only and also detect transitions from single to dual-stack or vice-versa. This information can be used by the DHCPv6 Relay Agent to influence the DHCPv6 Server to send appropriately prioritized DNS Servers to the client. The DHCPv6 server can implement the following based on connectivity information received from the relay agent. IPv6-only transition to Dual-Stack: In case a host is IPv6-only, it is provided with a DNS64 server. When transitioning to dual-stack, an IPv4 DNS server is assigned as a consequence of obtaining an IPv4 Address. The DHCPv6 Relay Agent can detect this and send a RECONFIGURE_REQUEST message to the DHCPv6 Server indicating that the host needs to be provided with a regular DNS server. In lieu of this mechanism, the host would continue to use the DNS64 server until the host stack reinitializes. Dual-Stack to IPv6-only: In case a host is dual-stack, it is provided with a regular DNS server followed by DNS64 server. When transitioning to IPv6-only, the DHCPv6 Relay Agent can detect this change and send a RECONFIGURE_REQUEST message to the DHCPv6 server indicating that the host needs to be assigned a DNS64 server only. In lieu of this mechanism, the host would continue to use the regular DNS Server which is inaccessible and eventually time out to fail over to the DNS64 Server. The host will take additional time to fully initialize causing delays in connection.

The option () includes an 8-bit status code that indicates specific host connectivity characteristics. The option can be included by a DHCPv6 Relay Agent in RELAY-FORW and RECONFIGURE-REQUEST.

IPv4_TO_DUAL_STACK: Host is transitioning from IPv4-Only to Dual-Stack mode. IPv6_TO_DUAL_STACK: Host is transitioning from IPv6-Only to Dual-Stack mode. DUAL_STACK_TO_IPv4: Host is transitioning from Dual-Stack to IPv4-Only mode. DUAL_STACK_TO_IPv6: Host is transitioning from Dual-Stack to IPv6-Only mode.

DHCPv6 relay agents that implement this specification MUST be configurable for tracking host connectivity and inserting the OPTION_HOST_CONNECTIVITY option in RELAY-FORW and RECONFIGURE-REQUEST messages. To be able to notify details of hosts' connectivity, a relay agent must be able to track host connectivity. A Relay Agent can detect host connectivity type using mechanisms discussed in . The Relay Agent then includes this information in the appropriate DHCPv6 message. Relay agents need to maintain connectivity state of each host it can track. This ensures that notifications to the DHCPv6 server, especially DHCPv6 RECONFIGURE_REQUEST, are accurately sent when there is a change in status. If a relay agent loses state due to some reason (e.g., during restart events), it will build state again using the mechanisms described in and then send appropriate notifications to the server. Such notifications are redundant and a DHCPv6 Server can choose to ignore such redundant notifications from the relay agent. Redundant notifications are also possible when relay agents are deployed in fault tolerant mode.

DHCPv6 relay agents that implement this specification MAY include the option OPTION_HOST_CONNECTIVITY in the RELAY_FORW to indicate status of host connectivity.

DHCPv6 relay agents that implement this specification MUST be configurable for sending the RECONFIGURE_REQUEST message. The relay agent generates a Reconfigure-Request anytime status of host connectivity changes by including OPTION_HOST_CONNECTIVITY in the request.

A DHCPv6 Server that supports OPTION_HOST_CONNECTIVITY may either have specific configuration or built-in logic to process information available in the option and send configuration parameters in DHCPv6 responses. How the server consumes and acts on the information obtained in the option are outside the scope of this document. The DHCPv6 server may use this connectivity information, if available, in addition to other relay agent option data, other options included in the DHCPv6 client messages, server configuration, and physical network topology information in order to assign appropriate configuration to the client. The server MUST ignore the option if it doesn't recognize the status in the OPTION_HOST_CONNECTIVITY option. The server SHOULD maintain the latest status received from the relay agent. The server can use this state to match against subsequent notifications and only further process if there is change in status. A relay agent could, for reasons such as restart, fault-tolerant mode etc, send redundant notifications and matching of status at the server will avoid unnecessary processing and message exchanges.

Upon receiving a RELAY-FORW message containing OPTION_HOST_CONNECTIVITY, the server can send appropriate configuration in the RELAY-REPLY response. The server MUST NOT return this option in a RELAY-REPLY message.

Upon receiving a RECONIFURE-REQUEST message containing an OPTION_HOST_CONNECTIVITY option, the server MUST follow the mechanism described in to create and send Reconfigure message. The server MUST NOT return this option in a RECONFIGURE-REPLY message.

Relay Agents can actively keep track of all IPv4/IPv6 addresses and associated lease times assigned to hosts via the respective DHCP servers. Relay Agents can therefore detect transitions from single to dual-stack and vice-versa efficiently. In addition to this technique, relay agents closest to the client can detect transitions using snooping mechanisms. Network devices today use mechanisms such as ARP and NDP snooping (bindings learnt by snooping all NDP traffic, NS, NA, RS, RA) to determine host characteristics such as IPv4/IPv6 - MAC - DUID bindings. IPv4/IPv6 and MAC counters are also used to determine host liveliness. First hop devices that implement first hop security also track IP address bindings and determine binding updates such as temporary addresses, deprecated addresses, etc. Existing work such as and also aim to active current host bindings, all of which can be leveraged to track host addresses. These mechanisms help determine if a particular IP address family is inactive, has reverted to using a single stack even though it initially had dual-stack capabilities and detect active dual-stack usage after long periods of single-stack activity. Other techniques to track host connectivity can be envisaged. It is out of scope of this document to provide an exhaustive list of host tracking techniques.

This document describes an application of the mechanism specified in . Host tracking mechanisms MUST be reliable. If a relay is compromised, it may be used to force an uncompromised server abuse clients by triggering repetitive reconfigurations. Security considerations described in are applicable to this mechanism.

IANA is requested to assign the following new DHCPv6 Option Code in the registry maintained in http://www.iana.org/assignments/ dhcpv6-parameters: OPTION_HOST_CONNECTIVITY