From HiPBX Wiki

One of the driving ideas behind this project is that there's nothing stopping a VoIP system being MORE reliable than a legacy phone system.

That does mean, however, that you can't just throw a big pile of hardware together and expect it to work. You do need to think about things a bit. I've put together some examples of how to cable up a reliable system to avoid everything but catastrophic damage to the physical location of your servers - and you can fix that by seperating them.

You will see that everything has a feed from the UPS and from the GPO. This ensures that even if the UPS fails, you can still swap it out without an outage. And when the GPO goes out, everything will be running on one PSU, being fed soley from the UPS.

When you are organising the UPS, ask the client 'How long do you expect your phone system to run without power?'. Double that amount (unless it's already a reasonable amount - I would suggest speccing it for at least 2 hours, preferrably 4 hours). If they say 'indefinately', then you'll need to start organsing generators and other fun stuff. This is what's known as a 'solved problem' and just needs money thrown at it to make it work.