Here is the strongest approach to remote computer management that I can think of:
- Use a random number generator (RNG) device (with no networking capability) to generate a private key.
- Physically connect this device to the master computer. Move the private key onto the master and wipe it from the RNG device.
- On the master, derive the public key from the private key. Copy the public key onto another device with no networking capability. Let's call it a "public-key" device or PK device.
- For each new slave computer in the fleet, perform this sequence of operations:
-- Use an RNG device to generate a private key.
-- Physically connect the RNG device to the new slave computer. Move the private key onto the slave and wipe it from the RNG device.
-- Physically connect the PK device to the new slave computer. Copy the master's public key onto the slave computer.
- The slave computer can now use public-private key cryptography to authenticate all commands from the master computer.

Note: The approach I have described is focused on protection against a remote attacker. An attacker with physical access to the slave computer could use another PK device to change the "master public key" stored on the slave, thereby transferring control of the slave to a different master computer. I will simply note that physical access control is a much older problem, with different solutions, mostly involving sentries, weapons, and identity documents (although perhaps in the future the notification of a physical visit and the relevant identity documents/photographs may be sent ahead beforehand, signed by a trusted public key).

~ StJohn Piano
Author: StJohn Piano
Published: 2020-06-21
Datafeed Article 148
This article has been digitally signed by Edgecase Datafeed.
This article has been digitally signed by its author.
4 minute read (10 pages)
3 comments
Click here to ask the author a question.