What Prefix Length Should I Use for My Network?

Choosing the correct IPv6 prefix length is fundamental to building a scalable, manageable, and standards-compliant network. Unlike IPv4 where address conservation drove complex subnetting schemes, IPv6's vast address space enables a simplified, consistent approach focused on operational efficiency and aggregation.

Understanding CIDR Notation in IPv6

IPv6 uses CIDR (Classless Inter-Domain Routing) notation to express network prefixes, written as an address followed by a slash and prefix length: 2001:db8::/32. The prefix length indicates how many bits identify the network portion, with the remaining bits available for host addressing.

Critical Rule: Use Nibble Boundaries

Prefix lengths should always be multiples of 4 to align with hexadecimal nibble boundaries. This alignment simplifies DNS reverse zone delegation and routing operations. Valid prefix lengths include /48, /52, /56, /60, and /64.

Non-aligned prefixes (like /63 or /65) create operational complexity in DNS delegation, routing table management, and troubleshooting. While technically possible, they violate best practices and should be avoided.

Common Prefix Lengths and Their Uses

/48 - The Enterprise Standard

Allocation: 65,536 /64 subnets Typical Use: Enterprise networks, data centers, business customers

The /48 is the gold standard for enterprise IPv6 allocation. It provides:

Over 65,000 subnets for comprehensive network segmentation
Minimum prefix length accepted for global BGP routing
Sufficient space for hierarchical address planning
Room for decades of growth without renumbering

Example Planning:

Allocated: 2001:db8:1234::/48

Building A: 2001:db8:1234:0100::/56 (256 subnets)
  Floor 1:  2001:db8:1234:0101::/64 (VLAN 101)
  Floor 2:  2001:db8:1234:0102::/64 (VLAN 102)

Building B: 2001:db8:1234:0200::/56 (256 subnets)
  Floor 1:  2001:db8:1234:0201::/64 (VLAN 201)

BGP Routing Consideration: Carriers will not advertise prefixes longer than /48 in the global routing table. If you need provider-independent address space with BGP multihoming, you must obtain at least a /48 from your Regional Internet Registry (RIR).

For more information on obtaining IPv6 address space, see IPv6 Address Assignment: How to Get IPv6 Addresses from RIRs and ISPs.

/56 - Residential and Small Business

Allocation: 256 /64 subnets Typical Use: Residential broadband, small branch offices

The /56 has emerged as the recommended minimum for residential Internet customers and small businesses. ISPs following RIPE-690 and similar best practices typically delegate /56 prefixes via DHCPv6 Prefix Delegation.

Why /56?

Enables home network segmentation (IoT, guest, primary)
Supports multiple VLANs for advanced users
Conforms to IPv6 standards for customer subnetting
256 subnets is ample for any residential scenario

Residential Subnetting Example:

ISP Delegation: 2001:db8:abcd::/56

Primary LAN:    2001:db8:abcd:00::/64
Guest WiFi:     2001:db8:abcd:01::/64
IoT Devices:    2001:db8:abcd:02::/64
Lab/Testing:    2001:db8:abcd:03::/64
...
(252 subnets reserved for future use)

Recent measurement data shows /56 usage in approximately 20% of observed networks, with adoption growing as ISPs modernize their IPv6 deployments via DHCPv6 Prefix Delegation.

/60 - Constrained Scenarios

Allocation: 16 /64 subnets Typical Use: Cost-conscious ISP deployments, limited scenarios

Some ISPs allocate /60 prefixes to balance address space management with customer needs. While providing 16 subnets may seem adequate, this approach is increasingly discouraged by operational best practice documents.

Limitations:

Minimal room for network growth
Insufficient for complex home networks (VLANs, multiple sites)
Does not align with RIPE-690 recommendations

When /60 Might Be Acceptable:

Legacy infrastructure constraints
Transitional allocations with upgrade path to /56
Strictly controlled environments with known subnet requirements

Current data indicates /60 usage in approximately 21% of networks, though this is expected to decline as /56 becomes the new standard.

/64 - The Individual Subnet

Allocation: 18,446,744,073,709,551,616 addresses Typical Use: Single network segment, LAN, VLAN

The /64 is the fundamental building block of IPv6 networks and the only prefix length that should be used for individual subnets where end devices connect. This is not a guideline but a hard requirement defined by IETF standards.

Why /64 is Mandatory for LANs:

SLAAC (Stateless Address Autoconfiguration) requires /64
Modified EUI-64 addressing uses 64-bit interface identifiers
Privacy extensions and temporary addresses assume /64
Breaking this standard causes unpredictable device behavior

Critical Warning: Assigning only a /64 to customers (without delegation capability) does not conform to IPv6 standards and breaks functionality. Customers must receive a prefix they can subnet (minimum /60, recommended /56 or /48).

Observed Usage: The /64 remains the most common prefix in operational networks (57% of cases), reflecting its role as the standard subnet size.

/127 and /128 - Special Cases

Point-to-Point Links (/127): RFC 6164 recommends /127 for point-to-point links between routers to prevent certain security and operational issues inherent in /64 allocations on P2P circuits.

Example P2P Link:
Router A: 2001:db8:ffff:0001::0/127
Router B: 2001:db8:ffff:0001::1/127

Single Host Addressing (/128): Used for loopback interfaces, anycast addresses, and specific host routes where only one address is needed.

Loopback: 2001:db8:1234::1/128
Anycast:  2001:db8:1234::100/128

IANA and RIR Allocation Policies

Global Hierarchy

IPv6 address allocation follows a hierarchical model:

IANA → RIRs: IANA allocates minimum /12 blocks to Regional Internet Registries
RIRs → LIRs/ISPs: RIRs allocate prefixes (typically /32 to /29) to Local Internet Registries and ISPs
ISPs → Customers: ISPs delegate prefixes to end customers

RIR Policies and Eligibility

Each RIR operates under its own policies, but common patterns exist:

Minimum Allocation Sizes:

Small ISPs and enterprises can typically qualify for /48
Larger ISPs receive /32 or shorter prefixes
Allocations sized based on immediate need and projected growth

Key Policy Principle: Unlike IPv4's scarcity-driven conservation, IPv6 policies focus on aggregation efficiency rather than address preservation. Allocate generously to enable proper route aggregation.

Provider-Independent (PI) vs Provider-Assigned (PA)

Provider-Independent (/48 minimum):

Portable across ISPs
Required for BGP multihoming
Must be obtained directly from RIR
Requires technical justification and annual fees

Provider-Assigned (varies):

Delegated by your ISP
Not portable if you change providers
Renumbering required if switching ISPs
More cost-effective for single-homed networks

ISP Deployment Practices

Business Customers

Standard Practice: /48 delegation Rationale: Business networks require extensive subnetting for:

Multiple locations and buildings
Departmental network segmentation
DMZ and security zones
Guest networks
Infrastructure networks (management, storage, backup)
Future expansion

Residential Customers

Current Best Practice: /56 delegation via DHCPv6-PD Conservative Alternative: Reserve /48, initially assign first /56

Some ISPs adopt a hybrid approach:

Reserve a /48 per customer in planning systems
Initially delegate the first /56 via DHCPv6-PD
Expand to full /48 on request or business upgrade

This approach maintains future flexibility while managing DHCPv6 prefix pool size.

Persistent vs Non-Persistent Prefixes:

RIPE-690 recommends persistent prefix assignment where customers retain the same prefix across sessions. For details on best practices, see IPv6 Deployment Best Practices and Standards. Benefits include:

Stable DNS records
Consistent firewall rules
Predictable remote access
Better for IoT device management

Enterprise Network Planning

Hierarchical Address Architecture

IPv6's abundance enables geographic and functional hierarchy without the constraints of VLSM:

Enterprise /48: 2001:db8:1000::/48

Geographic Regions (/52):
├─ North America: 2001:db8:1000::/52
├─ Europe:        2001:db8:1100::/52
├─ Asia-Pacific:  2001:db8:1200::/52
└─ South America: 2001:db8:1300::/52

Per-Site Allocation (/56):
└─ New York Office: 2001:db8:1000::/56
   ├─ Building A: 2001:db8:1000:00::/60
   │  ├─ Floor 1 Data: 2001:db8:1000:00::/64
   │  ├─ Floor 1 Voice: 2001:db8:1000:01::/64
   │  └─ Floor 2 Data: 2001:db8:1000:02::/64
   └─ Building B: 2001:db8:1000:10::/60

Planning Worksheet

Step 1: Determine Current Requirements

Count existing subnets/VLANs
List planned expansion (3-5 years)
Consider geographic distribution
Account for special-purpose networks

Step 2: Calculate Minimum Prefix

1-16 subnets: /60 minimum (discouraged)
17-256 subnets: /56 minimum
257-65,536 subnets: /48 required
Need BGP routing: /48 required

Step 3: Add Growth Buffer

Multiply current needs by 4-10x
Round down to next nibble boundary
Prefer shorter prefixes for flexibility

Step 4: Document Allocation Scheme

Create addressing policy document
Define subnet ranges by function/location
Reserve blocks for future use
Document in network management system

Subnet Design Considerations

Aggregation Over Conservation

The paradigm shift from IPv4 to IPv6 planning:

IPv4 Mindset (Obsolete for IPv6):

Conserve every possible address
Use VLSM to minimize waste
Justify every subnet allocation
Complex, fragmented address plans

IPv6 Mindset (Correct Approach):

Allocate generously for aggregation
Use consistent prefix lengths
Simple, hierarchical addressing
Future-proof with excess capacity

Functional Segmentation

Allocate address blocks by function for operational clarity:

Infrastructure: 2001:db8:1000::/56
├─ Loopbacks:   2001:db8:1000:00::/64
├─ P2P Links:   2001:db8:1000:01::/64 (/127s carved out)
└─ Management:  2001:db8:1000:02::/64

User Networks:  2001:db8:1100::/56
├─ Employees:   2001:db8:1100:00::/64
├─ Contractors: 2001:db8:1100:01::/64
└─ Guest WiFi:  2001:db8:1100:02::/64

Servers:        2001:db8:1200::/56
├─ Web/App:     2001:db8:1200:00::/64
├─ Database:    2001:db8:1200:01::/64
└─ Storage:     2001:db8:1200:02::/64

DNS Reverse Zone Delegation

Nibble-aligned prefixes simplify DNS operations:

Forward: example.com
Reverse: 0.0.0.1.8.b.d.0.1.0.0.2.ip6.arpa. (/48)

Delegating /56 to branch:
0.0.1.0.0.0.1.8.b.d.0.1.0.0.2.ip6.arpa. → branch DNS

Non-nibble-aligned prefixes require complex RFC 2317-style delegation mechanisms, creating operational burden.

Best Practices and Recommendations

For Enterprises

Request a /48 minimum - Even if current needs are smaller, growth inevitably happens
Use /64 for all LANs - No exceptions; required for standards compliance
Use /127 for P2P links - Follows RFC 6164 security recommendations
Plan hierarchically - Geographic → functional → subnet
Document extensively - Future network engineers will thank you
Consider PI space for multi-homing - Provider independence requires RIR allocation

For ISPs

Delegate /56 to residential customers - Minimum recommended by RIPE-690
Delegate /48 to business customers - Standard enterprise requirement
Never delegate only /64 - Breaks IPv6 standards and customer functionality
Use persistent assignments - Stable prefixes improve customer experience
Implement DHCPv6-PD - Standard mechanism for prefix delegation
Plan for growth - Request adequate allocation from RIR upfront

For Home Users

Verify ISP delegation size - Check if you're receiving /56 or /60
Plan subnet segmentation - IoT, guest, primary networks
Use DHCPv6-PD on router - Automatic prefix delegation from ISP
Assign /64 per network - One /64 per VLAN/segment
Test your connectivity - Use tools like test-ipv6.run and our IPv6 Connectivity Test Guide to verify proper IPv6 configuration

Common Mistakes to Avoid

Never:

Assign only /64 to customers who need subnetting capability
Use non-nibble-aligned prefixes (/63, /65, etc.)
Apply IPv4 conservation thinking to IPv6
Assume current needs won't grow
Skip documentation of addressing scheme

Always:

Allocate hierarchically for aggregation
Use /64 for individual subnets where hosts connect
Consider DNS reverse delegation in planning
Leave room for 5-10x growth
Follow RIR and IETF best practices

Testing Your IPv6 Configuration

After implementing your IPv6 prefix allocation, verify your network's IPv6 connectivity:

Visit test-ipv6.run to test dual-stack operation
Verify both IPv4 and IPv6 connectivity work correctly
Check if your network prefers IPv6 when both protocols are available
Monitor latency differences between IPv4 and IPv6 paths

Proper prefix allocation is the foundation of a well-designed IPv6 network. Following these guidelines ensures your network is scalable, manageable, and compliant with modern IPv6 standards.