Importing Devices
If you already have a list of what’s on your network, don’t redraw it. File → Import devices (CSV / Nmap / draw.io)… turns it into devices on the canvas, grouped into network segments.
Files are read in your browser. The file itself is never uploaded — only the diagram you go on to save.
From an asset list (CSV)
Section titled “From an asset list (CSV)”Export your inventory from wherever it lives (a spreadsheet, a CMDB, an RMM tool) as CSV, with one device per row and a header row. Schema recognizes the usual column names:
| For this | Use a column called (any of) |
|---|---|
| Name | Name, Device, Device Name, Asset, Label |
| Hostname | Hostname, Host, FQDN, Computer Name |
| IP address | IP, IP Address, IPv4, Address, Mgmt IP |
| Device type | Type, Device Type, Category |
| Network / segment | Network, Segment, Zone, Subnet, Group |
| VLAN | VLAN, VLAN ID |
| Subnet | CIDR, Prefix |
| Role, OS, Environment, Criticality, Owner, Location, Vendor, Serial, MAC, URL, Notes | The obvious names — e.g. Operating System, Env, Team, Site, Manufacturer, Asset Tag, Description |
Only a name, hostname or IP column is required. Columns Schema doesn’t recognize are skipped, and listed so you know. Comma, semicolon and tab-separated files all work.
Type is matched against Schema’s icon names first (“Firewall”, “Switch”, “Database”), then guessed from keywords in the type, role, OS and hostname (fw-edge-01 becomes a firewall; “Windows 11” becomes a workstation). Anything unrecognized becomes a server — change it afterwards from the Properties panel.
From an Nmap scan
Section titled “From an Nmap scan”Run a scan with XML output and import the file:
nmap -sV -O -oX scan.xml 10.0.0.0/24Every host that was up becomes a device, with its IP, hostname, MAC address and vendor, detected OS, and a note listing its open ports. Its type is taken from the OS fingerprint where there is one (a FortiGate is a firewall even though it serves a web page), otherwise from its open ports (5432 is a database; 9100 is a printer).
A scan can’t see how hosts are wired together, so no connections are drawn — add those yourself.
How devices are grouped
Section titled “How devices are grouped”For an asset list or an Nmap scan:
- If a row has a network/segment value, it goes in a segment of that name.
- Otherwise, if it has an IPv4 address, it goes in a segment for its
/24— named for that subnet, and with the subnet filled in, so network checks work immediately. - Otherwise it’s placed on the canvas ungrouped.
Imported content lands to the right of anything already on the canvas.
From a draw.io diagram
Section titled “From a draw.io diagram”The draw.io diagram option redraws a diagram made in draw.io (diagrams.net) — one you drew by hand, or one a discovery tool generated. Choose the .drawio file (or .xml; they’re the same thing). It’s a separate option because the file is already a picture rather than a list: devices stay where they were drawn, and the lines between them come across as connections.
| In the draw.io file | Becomes |
|---|---|
| A shape with a label, an icon from a shape library, or anything with a line attached | A device |
| A labelled box or swimlane with devices in it | A network segment — or a zone, if it holds other segments |
| A line between two devices | A connection |
| Free text, titles, the legend, unlabelled decoration | Left out |
Device details are read from wherever the diagram keeps them:
- The first line of the label is the name. An IP address on a later line is the device’s address.
- Lines in the shape’s tooltip written as
Key: valuefill in the matching field, using the same names as a CSV’s columns (Mgmt IP: 10.0.0.2,Model: Catalyst 9300). A bare hostname is taken as the hostname. Anything else is kept as notes. - The shape’s custom properties (Edit Data in draw.io) are read the same way. Properties that match no field are kept as the device’s custom fields.
- The device type comes from the icon — a Cisco, AWS or network-library shape is matched to Schema’s nearest icon — then from the same keyword guessing as an asset list.
Connections keep their label. Labels drawn at the two ends of a line are read as the port at each end and kept in the connection’s notes, as Ports: sw-dist-01 Te1/0/1 ↔ sw-acc-07 Te1/1/7 — select the connection to see them. The line’s tooltip is added to its notes too. Arrowheads and straight/right-angled routing carry over. Schema keeps one connection between any two devices, so if the diagram draws several lines between the same pair, the first becomes the connection and the others are listed in its notes.
Colours don’t come across — in Schema, a connection’s colour says whether it crosses a network boundary. But if the diagram has a legend (a small colour swatch with text beside it), Schema uses it to keep what the colours meant: a device drawn in a legend colour gets that text as its role, and a connection gets it as part of its label. So a diagram that colours its 10G links blue arrives with those connections labelled 10G.
A draw.io icon is usually smaller than a Schema device. If devices are drawn too close together to fit, the whole diagram is spread out evenly, just enough that nothing overlaps.
Every page of the file is read, and the pages are placed side by side. The import window lists what was left out.
Importing again
Section titled “Importing again”Importing uses the same matching as Merging Diagrams: a device already on the canvas (matched by IP address, then by name) is left where it is, and gains any details the new file has that it was missing. So importing this month’s export on top of last month’s diagram adds the new devices and fills in blanks — without duplicating anything or disturbing your layout. In a draw.io import, a connection between two devices that are already connected is left as it is.
An import can be undone with Ctrl+Z. Up to 2,000 devices can be imported at a time.