NAT Traversal (NAT-T) in IPsec
What You'll Learn
You'll understand how NAT-T solves the problem of IPsec traffic breaking when passing through NAT devices by encapsulating ESP packets in UDP.
The NAT Problem for IPsec
IPsec was designed assuming end-to-end connectivity with public IP addresses. But Network Address Translation (NAT) breaks IPsec in two ways:
- ESP header alteration: NAT devices modify IP headers (addresses and checksums). ESP's integrity checks detect this as tampering and drop packets.
- No transport-layer ports: ESP is IP protocol 50—it has no TCP/UDP port numbers for NAT to use for mapping multiple internal hosts.
Think of it like trying to mail a tamper-evident envelope through a postal system that rewrites addresses on sealed packages. The recipient's tamper seal check fails every time.
How NAT-T Fixes This
NAT Traversal (NAT-T) wraps ESP packets inside UDP datagrams, typically using UDP port 4500. This gives NAT devices the port numbers they need, while keeping the ESP payload protected inside.
The process:
- During IKE negotiation, endpoints detect NAT presence by comparing internal/external IP addresses
- If NAT is detected, both sides switch from ESP (protocol 50) to UDP encapsulation
- ESP packets get wrapped:
IP header | UDP header (port 4500) | ESP packet - NAT treats it like any UDP traffic, translating addresses and ports
- The receiving endpoint strips the UDP wrapper and processes ESP normally
It's like putting your tamper-evident envelope inside a regular envelope that the postal system can handle—the outer wrapper gets rewritten, but the inner protected content remains intact.
Automatic Detection
Modern IPsec implementations automatically detect NAT and enable NAT-T without manual configuration, making IPsec viable for remote workers behind home routers.
Key Takeaway: NAT-T encapsulates ESP inside UDP port 4500 to allow IPsec traffic to traverse NAT devices, solving both the port-mapping and integrity-check problems that would otherwise break IPsec connections.