That’s generally bad advice. Only some flaws get CVEs & livepatches. There are plenty of other kernel changes that bring stability, features, and fixes for less severe flaws, which can only be utilised with a reboot. Some systems have onerous uptime requirements and that trade-off is worth it. Most servers are better off rebooting as soon as practicable. If you cannot afford reboots, your problem is more likely insufficient redundancy.