Skip to main content
IT BIBLESign in

Blog / Networking

Labeling Every Cable and Building a Network Binder You Can Trust

A labeling scheme that survives changes, a port map you can check against the live switch, and the one habit that keeps a binder honest: write the change down before you leave the room.

RackGhostDwayne
· 6 min read · Networking
Labeling Every Cable and Building a Network Binder You Can Trust

Photo by Jordan Harrison on Unsplash

I am rebuilding my home network documentation from scratch, and the first lesson is the same one I learned in my twenties: a label that is wrong is worse than no label. With no label you go and look. With a wrong label you trust it, and you unplug the wrong thing.

Here is the scheme I settled on and why. None of it is clever, which is the point.

What goes on the label

The rule is simple. Label things with facts that do not change, and keep the facts that do change in the binder.

An IP address changes. A hostname changes. A VLAN assignment changes the first time you get an idea on a Saturday. The physical location of a wall jack, the number of a patch panel port, and which end of the cable you are holding do not change unless you pull the cable out of the wall.

So each cable gets an ID that points at its two ends, and nothing else. Mine look like this:

PP1-14 <-> SW1-p07

PP1-14 is patch panel 1, port 14. SW1-p07 is switch 1, port 7. The same text goes on both ends of the patch cable, so whoever picks up either end can read where the other one goes. The in-wall run behind the panel gets its own ID that matches the jack, for example BR2-A (basement room 2, jack A), printed on the keystone and on the wall plate.

Put the label within a couple of inches of the connector. A label in the middle of a six-foot patch cable is useless once the cable is in a bundle. Use a wrap-around label with a clear tail, or a flag label, so it can be read from the side without twisting the cable. I print mine. A handwritten label is fine for a week and illegible in a year.

The binder pages

Mine has five sections. A section is only worth having if I will open it when something breaks.

  1. One-page diagram. Boxes for the firewall, switches, and the server, lines for the uplinks, and the VLAN numbers beside each trunk. If it does not fit on one page, it is two diagrams, and I have drawn the wrong one.
  2. A port map for each switch and each patch panel.
  3. An addressing table: every subnet, its VLAN, its gateway, and the DHCP range.
  4. A change log, newest entry at the top.
  5. A recovery page: how to reach the firewall and the hypervisor when the network is down, and where the config backups are stored.

A port map is a plain table. Mine looks like this:

SW1 (24-port, rack, U12)
port  cable ID        far end              mode      VLAN
p01   PP1-01 <-> p01  uplink to firewall   trunk     10,20,30
p02   PP1-02 <-> p02  Proxmox host nic0    trunk     10,30
p07   PP1-14 <-> p07  BR2-A (desk)         access    10
p08   (empty)         -                    disabled  -

The empty row matters. A port marked "disabled, empty" tells me I checked it. A blank row tells me I forgot.

Checking the binder against the live gear

A binder written from memory is a guess. Once the tables are filled in, check them against what the equipment says, and do it port by port.

On each switch, pull the link state and the MAC address table. The command names differ by vendor, so use your own switch's equivalent of these two:

show interfaces status
show mac address-table

If the binder says p07 is a desk on VLAN 10 and the switch says p07 is down, someone moved a cable or I mislabeled a jack. If the MAC table shows three addresses on a port I recorded as a single desktop, there is a small switch or an access point hanging off it that I forgot.

If your switches and hosts speak LLDP, the neighbor table does a lot of the work for you. On a Linux host with the lldpd package installed, this lists what the host sees on the wire:

lldpctl

Not every device sends LLDP, and consumer gear mostly does not, so treat a silent port as unknown, not as empty. When in doubt, do the old method. Pull the cable, watch which port drops on the switch, plug it back, and write down what you saw. Two minutes per port, and it is the only check that cannot lie.

For the Linux side of the addressing table, these are the commands that give you the facts to copy down. If you want to practice them first on a throwaway machine, this lab covers address, reachability, and name resolution:

There you will find this machine's address, reach a web server by address and by name, and see what a failure looks like.

ip -br addr
ip route
resolvectl status

Those three show the address, the default gateway, and the DNS servers the host is actually using. The last one assumes systemd-resolved. If your distribution does not use it, look at /etc/resolv.conf instead.

The part that keeps it honest

The diagram is easy to draw once. The hard part is year two, when the binder and the rack have drifted apart.

My rule is that the change is not finished until it is written down. I move a cable, I put a pencil mark in the change log before I close the rack door: date, what I changed, why, and how to undo it. Four lines. If I wait until later, "later" does not come, and I know that about myself.

Two habits help.

  • Keep a pencil and a pad at the rack, not in the house. A note made upstairs is a note made from memory.
  • Once a year, pick a few ports at random and check them against the binder. If two of ten are wrong, the binder is not trustworthy and you should do the whole walk again. If all ten match, you can sleep.

A word on where this lives. A paper binder has one weakness: it is in one place. I keep a copy of the addressing table and port maps as plain text files in the same backup that holds my firewall config. Paper for the day the network is down, text for the day the binder is wet. I would not trust either one alone.

One caution. Do not put passwords in the binder. Put the location of the password vault and how to open it, and make sure someone besides you knows that location. Writing secrets on paper in a basement is a risk, and I would rather not talk myself into it.

This costs time up front. Labeling a 24-port panel properly takes me most of an afternoon, mostly the walking and the verifying. It also costs a label printer and a roll of labels, which I consider cheap against a single evening of tracing a cable by feel.

Your next step is small. Pick one switch, make a port map for it, and verify every row by pulling the cable. Then tell me what you found that did not match. There is almost always something.

Dwayne

networkingcablingdocumentation
Share

Have a question about this guide?

Create a free account to ask the community and join the discussion. Members who run the real thing answer every day.