You will never ever get a 100% redudant Data Center. It is impossible. As the article says a natural disaster could kill a whole data center. The only solution I could think of is to have 2 data centers in totally different areas of the world and to rsync each and every server BUT again this is still not 100% redundant as something can go wrong. The chances are less but you can not provide 100% redudandant service
To add to tris's post, you cannot eliminate all risk anywhere, you just minimise it. For example you could a disaster movie situation where a comet destroys everything for example, extremely unlikely but still possible. So just minimise the risk as far as reasonable possible.
TOS and AUP very important unless you wanna run into problems - especially where there is a sue culture. (UK hosts and datacenter's want to particulary careful - courts don't like company's been negiligent like that) I also supect the same may apply to US and some eu courts.
Well definetly for UK courts they are tough on businesses.
BACK ON TOPIC:
As jatos sent you can be safer but you can never be totally safe. Even with remote backups a hacker can get into them and delete them. There are some methods which make it SOOO unlikely that anything will happen but the fact that it could still happen means it isnt 100% redundant.