[bob1029]

The article does touch on HSMs but might be missing the point of them?

> A compromised server no longer means a compromised key

Proper use of an HSM means that even the owner of the private key is not allowed to access it. You sign your messages within the secure context of the HSM. The key never leaves. It cannot become compromised if the system is configured correctly.

[tjoff]

You can't get the private key but you can sign with it, which is still plenty bad.

[bob1029]

The private key should be tightly scoped to its context of use. I would definitely agree with you if it's one key that rules the entire kingdom.

[tjoff]

Not sure I follow? Lets say it is limited to one use only, sign an app.

Since I've got control of the box I can now use it to sign any app. Isn't that bad enough?

[eamann]

You're correct there that proper use means even the owner can't access it. But in a single key scenario they can still act unilaterally. The advantage of TSS is the removal of that level of unilateral action.

[mhluongo]

Again and again, we've seen that HSMs aren't secure against physical access of the device.

[bob1029]

Can you point me to an example of a FIPS level 3+ certified device having its private keys compromised due to a defeat of the tamper resistant boundary?

[mhluongo]

Here are a couple examples of physical access leading to key extraction. You're welcome to be pedantic (those are side channel attacks, they don't defeat the boundary!) but one way or another, physical access wins.

https://www.cl.cam.ac.uk/~rnc1/descrack/ https://ninjalab.io/eucleak/

[goodpoint]

No, if an HSM is compromised everything is lost.