Start from what runs and who uses it, not from the equipment list. Each workload is placed in a segment with a stated purpose: user, server, management, DMZ. What may talk to what is written down per segment, so the rules exist as a document before they exist as configuration.
The result is a plan that an auditor can read and an engineer can build from: the segments, the boundaries between them, the permitted flows across each boundary and the reason for every exception.
- Segments with a stated purpose and owner
- Permitted flows per boundary, written down
- Addressing and naming that survive growth
- The diagram kept current with the configuration

