Deploying professional software on a network with no internet
The practical steps for installing, licensing and updating software on an isolated or air-gapped network, including the parts that catch people out at rollout.
Installing software on an isolated network is not hard. It is just different, and the differences bite at rollout if nobody planned for them. This is the checklist we would give a customer before the first install.
1. Confirm what "isolated" means in practice
Three very different environments get called air-gapped:
- No internet, but internal network exists. You can move files between machines on the LAN, and you have a file server.
- No external network at all. Media transfers only, and possibly an approval process for each one.
- Physically separated with controlled media. There is a documented transfer procedure and a person who owns it.
The deployment plan differs for each. Establish which one you are in before anything else, in writing, and get the name of whoever approves a media transfer.
2. Get the installer onto the network legitimately
This is where projects stall. Someone assumes a USB stick can be plugged in, and it cannot. Plan the transfer with the security team up front: what media is approved, who scans it, what the documentation requirement is. Build that lead time into the schedule.
Prefer a portable build — a folder you unzip and run — over an installer that needs to write to system locations and call home. It avoids admin-rights arguments and leaves nothing behind to uninstall.
3. Licence before you unplug, or use an unbound edition
Device-bound activation normally needs one exchange: the app shows a device ID, the vendor signs a code, the code goes back in. On an isolated network that exchange happens over the media transfer process, which means it needs to be planned like any other transfer.
Where that is unacceptable, an edition with an unlimited built-in licence removes the exchange entirely. It costs more, and it is the right purchase for an environment where every transfer is a week of approvals.
4. Test on one machine, not the fleet
Install on a single representative machine and use it for real work for a week. Check the things that only show up in use: file sizes, printing, the paths your users actually save to, whether the app writes anything to a temp directory that is cleared on reboot.
5. Plan the update path before you need it
An offline network does not get patches for free. Decide now:
- How will a security fix reach these machines?
- Who decides when a version is deployed, and to which machines?
- Can the application run side by side with the previous version during a rollout?
If the answer to the second is "nobody knows", that is the finding. Write it down now rather than during an incident.
6. Document the state
Keep a file listing, per machine: what is installed, which version, which licence, and when it was last verified. On a connected network the vendor's dashboard holds this. On an isolated one, this document is the only inventory that exists, and it is the first thing anyone asks for.
The failure mode to avoid
The characteristic failure of offline deployment is not a technical one. It is that the person who understood the transfer process leaves, and six months later nobody can install an update. Write the procedure down, name two owners, and test the procedure once a year by actually doing a transfer.