nftables Rule Syntax and Families
You'll learn how to write nftables rules across different protocol families and understand modern matching expressions.
The Core Concept
Unlike iptables (which required separate commands for IPv4, IPv6, and other protocols), nftables introduces families that let you organize rules by network layer protocol in a unified framework. Each family operates on a specific protocol type, and you write rules using a more readable, consistent syntax across all of them.
The Five Main Families
ip– IPv4 only (like old iptables)ip6– IPv6 only (like old ip6tables)inet– Both IPv4 and IPv6 together (modern dual-stack)arp– ARP protocol (like old arptables)bridge– Bridged traffic (like old ebtables)
Think of families as different "languages" your firewall speaks. The inet family is like being bilingual—it understands both IPv4 and IPv6 in one ruleset.
Modern Rule Syntax
nftables uses a cleaner syntax:
nft add rule inet filter input ip saddr 192.168.1.0/24 tcp dport 22 accept
This rule uses the inet family (dual-stack), matches IPv4 source addresses from a subnet, checks TCP destination port 22, and accepts the packet.
Key improvements:
- Protocol-specific matching is explicit:
ip saddr(IPv4) vsip6 saddr(IPv6) - More natural language:
tcp dport 22instead of--dport 22 - You can use
inetfor most rules and let nftables handle both IP versions
Choosing the Right Family
Use inet for most firewall rules today—it's simpler and handles modern dual-stack networks. Use ip or ip6 only when you need protocol-specific features. Reserve arp and bridge for specialized layer-2 security.
Key Takeaway: nftables organizes rules by protocol family (ip, ip6, inet, arp, bridge), with inet being the modern choice for writing single rules that handle both IPv4 and IPv6 traffic using cleaner, more consistent syntax.