Originally Posted by raymor

On the "1 hour hardware replacements", if a drive fails it may take several more hours to
copy the data or restore from back up, so you'll want at least RAID 1,5, or 10 (not RAID 0).

Also if you're having a LOT of hardware problems, be aware that with SOME web hosts,
every problem is reported as a "hardware problem".

Basically, there are two ways to go with this. Live local fail over and remote backup fail over. Each has advantages and disadvantages. It sounds like you may want both - local for
nearly instant fail over and remote for when someone says "oops".

local failover
protects ONLY from actual hardware failure - not mistakes, not hackers, not DC issues, etc.
1-5 minute switchover
cost roughly double the normal single server cost + $750 set up and extra hardware

remote backup failover
protects from everything - such as the Planet fire which took out 9,000 servers, the FBI raids where they've confiscated entire data centers due to one customer, etc., mistakes, hackers, etc.
cost is $24 / month
switch over roughly 10-20 minutes