IP ADDRESSING

IPv6 Addressing: A Practical Operations Guide

How to read, compress, expand and plan IPv6 prefixes without getting lost in 128-bit notation.

IP addressing and CIDR illustration

The notation looks harder than the math

IPv6 uses 128-bit addresses written as eight 16-bit hexadecimal groups. A full address such as 2001:0db8:1234:5678:0000:0000:0000:0001 is long, so the notation allows leading zeros to be removed and one run of zero groups to be replaced with ::.

That example can be written as 2001:db8:1234:5678::1. The compressed and expanded forms identify the same address. Being able to move between them is useful when comparing device output, logs and documentation that use different formatting.

Prefix length still defines the network

Just like IPv4 CIDR, an IPv6 prefix length tells you how many high-order bits identify the network. A /64 leaves 64 bits for the interface portion. Many LAN designs use /64 because IPv6 neighbor discovery and address-autoconfiguration behavior commonly assume that boundary.

Other prefix lengths have valid uses. Loopbacks are often /128, routed customer allocations may be /48 or /56, and point-to-point infrastructure can follow an organization’s chosen standard. The important part is consistency with the feature set and design.

Do not bring IPv4 scarcity habits into IPv6

IPv6 planning is usually hierarchical rather than focused on squeezing every last address out of a block. Allocate on clean nibble boundaries where practical, keep sites and functions easy to summarize, and document the plan.

The goal is operational simplicity. A readable address plan can encode region, site or function in stable fields without creating an unnecessarily complex scheme that only its designer understands.

Watch link-local and global scope

Every IPv6-enabled interface can have a link-local address in fe80::/10 in addition to global or unique-local addresses. Routing protocols and next-hop displays may use link-local addresses, which can surprise engineers accustomed to IPv4.

Scope matters. A link-local address is meaningful only on the local link and may require an interface identifier when used in commands. Do not mistake it for a globally routed address.

Operational checks during deployment

Validate router advertisements, prefix length, default gateway behavior, DNS AAAA records, firewall policy, monitoring and path MTU. Dual-stack environments also need clear troubleshooting steps so engineers know whether a failure is IPv4-specific, IPv6-specific or application-related.

Treat IPv6 as a first-class routed service rather than an add-on. That includes logging, telemetry, ACLs, DDoS controls and change documentation.

← Back to all guidesBrowse network tools →
Use these guides as engineering references, not as a substitute for your network design standards, current vendor documentation or production change review.