4map6 Segments for IPv4 Service delivery over IPv6-only underlay networks

The document [I-D./draft-ietf-v6ops-framework-md-ipv6only-underlay] proposes a framework for deploying IPv6-only as the underlay in multi-domain networks. In this framework, each PE will be identified by at least one IPv6 mapping prefix, it will also have one or more associated IPv4 address blocks which are extracted from local IPv4 routing table or address pool. In addition, a specific data structure called address mapping rule is defined to express the mapping relationship between IPv4 address blocks and the IPv6 mapping prefix of the remote PE. With this design, if the mapping rules of the remote PE are obtained by the ingress PE, the mapping rule will give the forwarding guidance of IPv4 packet delivery in the IPv6-only network when the destination address of the IPv4 packet matches its IPv4 address block, the ingress PE will use the mapping rule to generate corresponding IPv6 source and destination addresses from its IPv4 source and destination addresses, and then generate a new IPv6 packet. This mapping-based conversion can also work in SRv6 networks. SRv6 defines packet processing in IPv6 network as a list of instructions, which are represented as 128-bit segments, often called Segment ID (SID). This document defines two new types of segments for Segment Routing, i.e., M46S and M46D segments, which are mapping rules based IPv4/IPv6 conversion function running in PE nodes. In multi-domain IPv6-only networks, the M46S and M46D segments at the ingress node can convert IPv4 packets into IPv6 ones by stateless encapsulation or translation, another M46S and M46D segments at the egress node can restore them to IPv4 packets after being transmitted in IPv6-only networks.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.

SRv6 SID has the same format as a 128-bit IPv6 address, As per section 3 of , a SID consists of LOC:FUNCT:ARG, where a locator (LOC) is encoded in the L most significant bits of the SID, followed by F bits of function (FUNCT) and A bits of arguments (ARG).

The LOC identifies the node which instantiates the SID and will lead the packets to the node. The FUNCT is an opaque identification of behavior bound to the SID. ARG field provides additional information for its processing. As new types of SID, M46S and M46D segments will follow the format of a general SID. Furthermore, several information items specific to stateless address mapping and packet conversion are carried in the relevant fields of the M46S SID and M46D SID, as below, • The LOC field is a prefix allocated by operators to identify the node which instantiates the M46S/M46D SID. • The FUNCT field identifies the behavior bound to the M46S/M46D SID, the behavior will be defined in Section 4. • The ARG field contains the IPv4 address prefix associated with the PE node. Since the IPv4 address prefix requires a maximum of 32 bits of a SID, the value of L+F should be less than or equal to 96. For the PE node which instantiates M46S/M46D SID, its IPv6 mapping prefix for IPv4 delivery corresponds to the combination of LOC and FUNCT fields. Generally, the number of SIDs that can be instantiated by both M46S SID and M46D SID is equal to the number of associated IPv4 address blocks. When an operator instantiates a M46S/M46D SID at an edge node, they specify a SID value LOC:FUNCT:ARG and the behavior bound to the SID, An SRv6 endpoint behavior may require additional information for its processing (e.g. tunnel or translation). This information may be carried in the control plane, which is out of the scope of this draft.

Generally, the M46S/M46D SID nodes run in pairs. For a specific data flow, one node is ingress PE, denoted by PE1, and the other is egress PE, denoted by PE2. All nodes located at the edge of the network need to announce their M46D SIDs to other nodes. PE2 announces its ability to other nodes in the format of M46D SID in the SRv6 domain at the control plane. When PE1 receives the M46D SID announced by PE2, it will bestored in the local database. This database can store multiple entries, and each entry includes IPv4 address block, Locator, Function, device capability and other information corresponding to the SID, The IPv6 mapping prefix, i.e. Locator and Function, represents the egress of the packet whose destination address is the IPv4 address block. When PE1 receives an IPv4 packet, it uses the destination IPv4 address block to look for the corresponding IPv4 address block entry in the local database. If a matching IPv4 address block entry is found, the corresponding IPv6 mapping prefix will splice the IPv4 destination address to generate a M46D SID, and the 32-bit IPv4 address is placed in the ARG field. Following the general SRv6 procedure, the SID is programmed into SRH. After that, the IPv4 destination address needs to be spliced with the corresponding IPv6 mapping prefix, and the M46S SID behavior identifier is placed in the FUNCT field to generate the M46S SID. The 32-bit IPv4 source address in the data packet is placed in the ARG field. Following the general SRv6 procedure, M46S SID is programmed into SRH. The newly generated packet with the SRH is sent into the IPv6-only network for further delivery. In this case, the source address SA of the IPv6 packet is the PE1 local address, and the destination address DA is taken from SRH according to the SRv6 procedure. The process is shown in the figure below,

When a new IPv6 packet arrives at PE2, PE2 parses its Locator part. If it matches the IPv6 mapping prefix instantiated by itself, it decapsulates the packet according to the instructions in Function, removes the SRH and IPv6 mapping prefix of the outer layer, and then forwards it to the next-hop according to the destination IPv4 address carried in the ARG field. When a new IPv6 packet arrives at PE2, PE2 resolves the LOC part of SRH in this packet layer by layer. If it matches the IPv6 mapping prefix of its own instantiation, it decapsulates the packet according to the instructions in the FUNCT. Recognition to the FUNCT field M46S and M46D respectively, to extract the corresponding to the ARG field source IPv4 address and purpose to carry the IPv4 address forward IPv4 packet to the next-hop recovery.

In an SRv6 network environment, the M46S/M46D SID needs to be advertised. The node advertises the M46S/M46D SID, B:N:FUNCT:ARG,through the control plane together with the SRv6 Endpoint Behavior codepoint identifying the behavior of the SID. Similar to other types of SIDs, the M46S/M46D SID can be distributed within and across domains via IGP and BGP or other approaches. The advertisement method and reachability calculation are specific to the chosen routing protocol. The distribution of the M46S/M46D Endpoint Behavior codepoint is left in future documents, e.g. by extending the SRv6 L3 Service TLV as defined in .

This document requests IANA to allocate the following codepoints in "SRv6 Endpoint Behaviors" sub-registry of "Segment Routing Parameters" top-level registry.

There is no additional security risk introduced by this design.