Depends on your threat model and actual realistic concerns.
Ultimately, if it comes down to it, there’s very little you can do that’s failsafe and 100% guaranteed: the provider has access to your disk, all data in your instances RAM (including encryption keys), and can watch your processes execute in real time and see even the specific instructions your vCPU is executing.
Don’t put illegal shit on hardware you do not physically own and have physical control over, and encrypt everything else but like, if the value of your shit is high enough, you’re fucked if you’re using someone else’s computer.
I mean, if you’re doing to do illegal shit (eg journalism), it is best done on a VPS. This is what basically every military and cyber mercenary orgs do.
Just make sure you only log into the box over Tor, and configure it to only pass data out over Tor. Use it as a jump box. Even if they compromise it, it should tell them nothing about you useful for attribution
If you have confidential compute(Intel TDX or AMD SEV) available from the cloud provider then it is feasible, provided you trust the CPU manufacturer. They should provide capabilities that allow you to perform remote attestation and ensure the virtual machine is running in the protected mode. Hypervisors running these types of machines usually can’t access the memory of the virtual machine, not even the registers.
Of course, nothing guarantees there won’t be a vulnerability that breaks it(e.g. side channel attacks), but right now the technology exists where you can run software in the cloud, protected from the provider…
That’s neat. Looks like that’s more recent than my most recent experience doing cloud work, but fully support it.
When I left there was some anti-fraud work being discussed around snooping on everyone’s executing processes to see who is doing bad things, and alerting when and if certain binaries were detected, and that was a little… gross.
Though, regardless of technical fixes, you’re still at the mercy of politics, policies, and people not doing sketchy shit. And, I’d also bet that not all of the smaller cloud providers are going to be doing the work for TDX or SEV implementations, so there’s still a big leap of trust you need of your provider, regardless of the existence of tech that could mitigate those issues.
Depends on your threat model and actual realistic concerns.
Ultimately, if it comes down to it, there’s very little you can do that’s failsafe and 100% guaranteed: the provider has access to your disk, all data in your instances RAM (including encryption keys), and can watch your processes execute in real time and see even the specific instructions your vCPU is executing.
Don’t put illegal shit on hardware you do not physically own and have physical control over, and encrypt everything else but like, if the value of your shit is high enough, you’re fucked if you’re using someone else’s computer.
I like this answer
I mean, if you’re doing to do illegal shit (eg journalism), it is best done on a VPS. This is what basically every military and cyber mercenary orgs do.
Just make sure you only log into the box over Tor, and configure it to only pass data out over Tor. Use it as a jump box. Even if they compromise it, it should tell them nothing about you useful for attribution
If you have confidential compute(Intel TDX or AMD SEV) available from the cloud provider then it is feasible, provided you trust the CPU manufacturer. They should provide capabilities that allow you to perform remote attestation and ensure the virtual machine is running in the protected mode. Hypervisors running these types of machines usually can’t access the memory of the virtual machine, not even the registers.
Of course, nothing guarantees there won’t be a vulnerability that breaks it(e.g. side channel attacks), but right now the technology exists where you can run software in the cloud, protected from the provider…
That’s neat. Looks like that’s more recent than my most recent experience doing cloud work, but fully support it.
When I left there was some anti-fraud work being discussed around snooping on everyone’s executing processes to see who is doing bad things, and alerting when and if certain binaries were detected, and that was a little… gross.
Though, regardless of technical fixes, you’re still at the mercy of politics, policies, and people not doing sketchy shit. And, I’d also bet that not all of the smaller cloud providers are going to be doing the work for TDX or SEV implementations, so there’s still a big leap of trust you need of your provider, regardless of the existence of tech that could mitigate those issues.