Bottom line is this: there is absolutely no good reason for not doing rolling updates. Do a few and make sure they are ok. Keep rolling out in groups. This single approach alone would've meant that this event was of marginal impact to most of the public, as sysadmins would've had the opportunity to halt further updates and work on remediating their first group (typically non-critical servers). Rolling out to everything all at once is just bad practice, period.