Seems like an argument for a heterogeneous environment, perhaps a solid and secure Linux server to host important keys like that.
Comment on An angry admin shares the CrowdStrike outage experience
db0@lemmy.dbzer0.com 1 month ago
Pity the administrators who dutifully kept a list of those keys on a secure server share, only to find that the server is also now showing a screen of baleful blue.
Lol, can you imagine? It empathetically hurts me even thinking of this situation. Enter that brave hero who kept the fileshare decryption key in a local keepass :D
kescusay@lemmy.world 1 month ago
pearsaltchocolatebar@discuss.online 1 month ago
Linux can shit the bed too. You need to maintain a physical copy.
Voroxpete@sh.itjust.works 1 month ago
Their point is not that linux can’t fail, it’s that a mix of windows and linux is better than just one. That’s what “heterogeneous environment” means.
You should think of your network environment like an ecosystem; monocultures are vulnerable to systemic failure. Diverse ecosystems are more resilient.
gnutrino@programming.dev 1 month ago
Sure but the chances of your Windows and Linux machines shitting the bed at the same time is less than if everything is running Windows. It’s exactly the same reason you keep a physical copy (which after all can break/burn down etc.) - more baskets to spread your eggs across.
pearsaltchocolatebar@discuss.online 1 month ago
Very few businesses are going to spend the money running redundant infrastructure on two different operating systems. Most of them won’t even spend the money on a proper DR plan.
noobface@lemmy.world 1 month ago
Hey Ralph can you get that post-it from the bottom of your keyboard?
StaySquared@lemmy.world 1 month ago
CS did take down Linux a few years back… I forget the exact details.
Avatar_of_Self@lemmy.world 1 month ago
Yes, but has it taken both OS’ out at the same time? It hasn’t but it could happen, however, the chances are even less. There’s obvious risk mitigation in mixing vendors in infrastructure for both hardware and software in the enterprise.
If some critical services were lost in your enterprise last time until RH updated their kernel then you could have benefitted from running that service from Windows as well. Now the reverse is true. You could have another DC via Samba on Linux in your forest if you wanted to, in order to have an AD still for example. Same goes for file share servers, intermediary certificate servers (hopefully your Root CA is not always on the network) and pretty much most critical services.
Most enterprises run a lot of services off of a hypervisor and have overhead to scale (or they are already in a sinking ship), so you can just spin up VMs to do that. It isn’t as if it is unreasonably labor intensive compared to other similar risk mitigation implementations. Any sane CCB (obviously there are edge cases but we are talking in general here) will even let you get away without a vendor support contract for those, since they are just for emergency redundancy and not anywhere near critical unless the critical services have already shit the bed.
amanda@aggregatet.org 1 month ago
Sounds like we may have an easier conclusion to draw here
sugar_in_your_tea@sh.itjust.works 1 month ago
That’s why the 3-2-1 rule exists:
For something like keys, that means:
Any IT pro should be aware of this “rule.”
IphtashuFitz@lemmy.world 1 month ago
We have a cron job that once a quarter files a ticket with whoever is on-call that week to test all our documented emergency access procedures to ensure they’re all working, accessible, up-to-date etc.
Empricorn@feddit.nl 1 month ago
Are you hiring!?