Tools / IPv6 subnet calculator
IPv6 subnet calculator
Try 2001:db8::/32 or 2001:db8:1:2::/64. Calculated in your browser.
2001:db8::/32
documentation
IPv6 has no broadcast address, so every address in the range is usable.
Working with IPv4? Use the IPv4 subnet calculator.
How to plan IPv6 subnets
An IPv6 allocation is large enough that you plan by structure, not by counting hosts.
- Prefix length
- A /n prefix fixes the first n of 128 bits. A shorter prefix is a bigger block.
- The /64 rule
- Every network segment gets a /64. That is the unit you hand out, because address autoconfiguration needs 64 host bits.
- Typical allocations
- A site is commonly given a /48, which holds 65,536 subnets of /64, or a /56, which holds 256. Residential connections are commonly given a /56 or a /60.
- Compressed and expanded
- The expanded form writes all 32 hex digits. The compressed form drops leading zeros and shortens the longest run of zero groups to "::". Both are the same address.
- Splitting
- Split a /48 into /64s to give each segment its own network, or a /32 into /48s to give each site one.
Keep each level of the plan on a hex-digit boundary (/48, /52, /56, /60, /64). The addresses stay readable and the routes stay easy to summarise.
Questions
How many addresses are in an IPv6 prefix?
A /n prefix leaves 128 minus n host bits, so it holds 2 to the power of (128 minus n) addresses. A /64 holds about 18.4 quintillion; a /48 holds 65,536 of those /64 subnets. The calculator prints the exact figure.
Why is /64 the standard subnet size?
Address autoconfiguration (SLAAC) builds an address from a 64-bit interface identifier, so a local network needs a /64 for it to work. Smaller prefixes are used for point-to-point links (/127, RFC 6164) and loopbacks (/128), not for networks of hosts.
Is there a broadcast address in IPv6?
No. IPv6 replaced broadcast with multicast, so every address in a prefix is a usable address and there is nothing to subtract from the total, unlike an ordinary IPv4 subnet.
How do the compressed and expanded forms differ?
The expanded form writes all eight groups of four hex digits. The compressed form drops leading zeros and replaces the longest run of zero groups with "::", following RFC 5952: lower case, and the leftmost run when two are equal. Both name the same address.
What do the reserved ranges mean?
2001:db8::/32 is for documentation, fc00::/7 is unique local (the private range), fe80::/10 is link-local, ff00::/8 is multicast and 2000::/3 is global unicast. The Type field above shows which one your prefix falls in.
Is anything sent to a server?
The maths runs in your browser and nothing is stored. A link you copy carries the address in its URL, like any link. The same calculation is available as a free endpoint: GET /api/public/tools/subnet?cidr=2001:db8::/32&split=48. It is rate limited per address.
Looking up an address instead? Try What is my IP or the DNS lookup.