Are you implying they should somehow have perfect quality control? Generally the lost customer trust is enough of an incentive to not let things like this happen. Things slip through QA, and the only way to prevent that at least 99.99% of the time is to invest MASSIVE amounts of money that really aren’t justified for everything. Aviation does this, because there is significant risk of death if something goes wrong, so regulations force them to. Other industries arguably should (car manufacturers…), but a random security software? No.
Nothing is perfect, nothing is absolute, and yes that’s an oxymoron but you get the point. Anyway, there are ways to minimize risk
A/B testing
gradual roll out
monitored roll out
rollback
And not only on the side of Crowdstrike, there are things that can be done by their customers:
OS rollback from weekly or monthly snapshots of the boot drive or system drive (probably shouldn’t change that often)
if that isn’t possible with that OS, use another OS
automated deployment (again, probably possible to fallback to a last known good deployment)
investment in sysadmins
investment in security staff
Probably lots more, but I’m not a sysadmin. I bet you though, that the hospitals, rail, and other governmental institutions simply don’t have enough money to invest in that because of budget cuts and austerity measures. Some hospitals still have Windows XP running.
Companies and governments don’t think IT and security are important until they are. It’s not about creating a perfect system, it’s about creating a system that can bounce back quickly.
Yeah, this absolutely smells like a corporate culture issue, not a one off glitch in QC. Fuckups of this magnitude shouldn’t be possible without multiple failsafes breaking and people ignoring protocol. Not to say that “perfect storm” events don’t ever happen, but it seems like the less-likely possibility to me.
Are you implying they should somehow have perfect quality control? Generally the lost customer trust is enough of an incentive to not let things like this happen. Things slip through QA, and the only way to prevent that at least 99.99% of the time is to invest MASSIVE amounts of money that really aren’t justified for everything. Aviation does this, because there is significant risk of death if something goes wrong, so regulations force them to. Other industries arguably should (car manufacturers…), but a random security software? No.
Car manufacturers have ISO 26262 to regulate safety critical software development, whereas aviation is mainly based on RTCA DO178.
The concepts are pretty similar. Details differ.
Nothing is perfect, nothing is absolute, and yes that’s an oxymoron but you get the point. Anyway, there are ways to minimize risk
And not only on the side of Crowdstrike, there are things that can be done by their customers:
Probably lots more, but I’m not a sysadmin. I bet you though, that the hospitals, rail, and other governmental institutions simply don’t have enough money to invest in that because of budget cuts and austerity measures. Some hospitals still have Windows XP running.
Companies and governments don’t think IT and security are important until they are. It’s not about creating a perfect system, it’s about creating a system that can bounce back quickly.
Anti Commercial-AI license
Yeah, this absolutely smells like a corporate culture issue, not a one off glitch in QC. Fuckups of this magnitude shouldn’t be possible without multiple failsafes breaking and people ignoring protocol. Not to say that “perfect storm” events don’t ever happen, but it seems like the less-likely possibility to me.