I've said this before, and I will say it again: Massive single serveruptime is not your friend.
Repeat after me: Kernel updates are there for a reason (I get at least one per month).
That being said ... this is a unique way of going about it.
Linux admin hated downtime so much he schlepped a live UPS during office move