Seemingly pretty low. I work in a large org and getting to the point of a quick streamlined solution hinged on just a few people. Without those few, it would have been a 100% manual effort. Most didn’t even seem to know a good way to identify the issue at scale.<p>The solution didn’t bother with safe mode, it used a boot image that would search for the offending file on all the drives and deleted it, then rebooted back onto the normal boot drive.<p>With specialization, so few people know all the various pieces to be able to quickly solution something. Good generalists are often overlooked, but they are the saviors in these moments.
Oh, below 20%, I'd wager. The kinds of sysadmin skills people used to have that would allow that to happen (PXE, serial over LAN, etc) have atrophied quite a bit in the last ~decade. Maybe around 20% if you limit it to people whose titles actually are "System Administrator" (which implies an older, steady-as-she-goes IT dept) instead of e.g. also lumping in DevOps or SRE or cloud people.<p>But probably above 1%, given that it's a serious enough tail risk that you might keep an old geezer around who remembers how to do it just in case something mission critical happens.
You need out of band remote management, basically a separate computer on the side that can do things to the main computer, or a read only OS with networking and management abilities that things fall back to. It’s definitely a thing in the datacenter world, and there are examples of cloud stuff. MDM solutions come to mind for mobile, and MacOS can do that too in a limited fashion.
I suspect that deploying a new image via PXE will be the fastest resolution.<p>Ideally one you just built without crowdstrike.<p>If you have important data trapped in userspace, print out some bitlocker keys and get in the car lmao.