NAT and CGNAT

What this topic solves

Address and port translation, state, source/destination NAT, CGNAT scaling and troubleshooting.

Follow state and packet path

Separate IP connectivity from the service itself. Capture the transaction and validate state, timers, relay/proxy functions and security policy.

Input → classification/state → decision → table → next-hop/output → counters

Design before configuration

• Define the failure domain and expected convergence behavior.
• Document addressing, identifiers, timers, MTU and policy boundaries.
• Prefer explicit, observable state over assumptions.
• Plan rollback and out-of-band access before risky changes.

Commands and verification

Command syntax varies by platform and software release. The examples below come from the existing portal content and must be validated against the target platform documentation before production use.

inside-local        inside-global
10.10.10.5:52144 -> 203.0.113.8:41022
                           |
Internet <-----------------+
retorno precisa casar com o mapping

LAN cliente: 192.168.1.0/24
      ↓ NAT no CPE
WAN/CPE: 100.64.12.34
      ↓ CGN do ISP
IPv4 público: 203.0.113.50
      ↓
Internet

Evidence-driven workflow

1. Define source, destination, time window and exact symptom.
2. Establish the expected healthy state and compare it with the observed state.
3. Inspect counters, tables, logs and packet captures before resets.
4. Test one hypothesis at a time and prefer reversible changes.
5. Repeat the original test after correction and check side effects.

Standards and RFCs applicable to the technology

Use the current standard/RFC and the documentation for the exact software release being operated. Vendor syntax is an implementation detail; protocol behavior should be understood independently.

Prove the behavior in a controlled topology

Build the smallest topology that reproduces the mechanism, capture control and data traffic, introduce one deliberate failure and diagnose it before applying the fix.

Related network topics