The /64 IPv6 Subnet: The Standard Network Size

The /64 IPv6 Subnet

A /64 subnet is the standard and recommended subnet size for IPv6 networks. The first 64 bits identify the network, and the remaining 64 bits identify the host (interface identifier).

Key Details

• Prefix Length: 64 bits
• Host Bits: 64
• Addresses per Subnet: 2^64 = 18,446,744,073,709,551,616 (approximately 18.4 quintillion)
• Purpose: One /64 per network segment (LAN, VLAN, point-to-point link)

Why Exactly /64?

The /64 boundary is not arbitrary — it is deeply embedded in IPv6's architecture:

1. SLAAC (Stateless Address Autoconfiguration): Hosts generate their own IPv6 address by combining the /64 network prefix with a 64-bit interface identifier (derived from their MAC address via EUI-64, or randomly generated for privacy). SLAAC requires exactly 64 host bits.

2. Neighbor Discovery Protocol (NDP): NDP, which replaces ARP in IPv6, is optimized for /64 subnets.

3. Simplified Routing: A consistent /64 boundary means routers always know where the network prefix ends and the host portion begins.

Address Generation Methods

On a /64 subnet, hosts can obtain addresses through:

• SLAAC — host auto-generates address (most common)
• DHCPv6 — centralized address assignment
• Static — manually configured
• Privacy Extensions (RFC 4941) — randomized interface ID that changes periodically

Can You Use Something Other Than /64?

Technically yes, but it is strongly discouraged. Using /65 or longer prefixes breaks SLAAC, and using shorter prefixes (like /48 directly on a link) wastes the subnet hierarchy. The only common exception is /127 for point-to-point router links (RFC 6164) and /128 for loopback addresses.

Addressing Capacity

With 2^64 addresses per subnet, address exhaustion within a single /64 is not a concern. Even if you assigned a million addresses per second, it would take over 500,000 years to exhaust a single /64 subnet.