Server power supply failures?
If all the servers have redundant PSUs (), you can plug just enough PSUs into the UPs and rest into the main powergrid. If you have 10 servers, with dual PSUs each (which makes 20 PSUs to plug into the power), you can simply plug only one of them into the UPS (so you would only take up 10 sockets in the UPS's PDB) - in case of power failure, the UPS will hold the servers up.
However, this does not cancel the previous comments about a UPS not being the ultimate answer to power failures.
Are you sure your UPS can handle to load of 4 additional servers?
Given the circumstances it would make the most sense to plug one PS into the UPS and the other PS to a surge-protected wall source. The servers can run with just power from the UPS to the one PS. The odds of a PS going bad at the same time as a power outage is very rare, though possible.
I agree with the previous answer. Also are all of you servers conencted to the ups correctly? For instance in a power outage, does the UPS software actually shutdown your machines gracefully. If not, they you are just as well off to plug the servers into your wall socket.
The moment the power goes out I predict that your UPS will run out of power before your servers can turn off.
In general, UPSes are used as a temporary stop-gap measure before a diesel generator can kick in.
It's been my experience that the UPS will fail far more often than the power supplies.
Also, the UPS is usually less reliable than the city power, and the power supplies are far more reliable than either.
That isn't to say that you shouldn't use a UPS, but in this situation, you really ought to have two UPSs, and make it so that a single UPS failure won't affect anything.
I've found that 2 power connections are most useful if you want to move a server from one outlet to another, or one part of the room to another, you can just walk it from outlet to outlet.
And, for goodness sake, monitor your UPSs. A UPS you don't pay attention to will interrupt service more often than anything else.