IPv6 Prefix
Leading portion of an IPv6 address that identifies a network, written in CIDR notation as `address/length`.
An IPv6 prefix is the leading n bits of an IPv6 address that identify a network, written in CIDR notation, e.g. 2001:db8:abcd::/48. The first n bits are the network portion; the remaining 128 − n bits address an interface inside that network. Because IPv6 drops the IPv4 concept of network and broadcast addresses, every bit pattern in the host portion is a usable address.
A handful of prefix lengths are conventional rather than arbitrary. /32 is the typical allocation from an RIR to an ISP. /48 is the canonical site allocation, leaving 16 bits of internal subnetting before the host boundary. /56 is common for residential customers, still leaving 8 bits of subnet room. /64 is the standard link prefix — almost every subnet on the Internet is a /64 because SLAAC and many Neighbor Discovery features assume a 64-bit interface identifier. Prefixes longer than /64 are valid but break SLAAC and should be reserved for special cases such as point-to-point links (/127) or loopbacks (/128).
Subnetting works as in IPv4: each extra bit of prefix length halves the network and doubles the number of subnets. Inside a /48 you have 65 536 /64 subnets to assign, which is why IPv6 site designs allocate per-VLAN, per-function or per-location subnets generously rather than packing addresses tightly.
Key Points
- Notation
address/length, length in bits (0–128) /32RIR allocation,/48site,/56residential,/64link/64is the practical smallest subnet — required by SLAAC- Longer than
/64allowed for point-to-point (/127) and loopbacks (/128) - No reserved network or broadcast address; the entire host portion is usable
- Each extra prefix bit halves the network and doubles the subnet count