UPS comms architecture

[Norman Dignard]

We're trying to better understand the functioning of our UPSes, management consoles and connect systems during and after a power failure.

We have a few proliants and well as risc servers connect either to its own ups of, in the case of the Proliants sharing the same ups. Some of the UPs have multiple serial ports connected to the servers with UPs sw installed. We do/did not configure a master managent console. Each server's UPS consule basicaly tells it to shoutdown after a few minutes on loss of power and to reboot on power restore. The UPSes also has other periferals (switches, disk arrays, etc) connected to it.

Wha I need clarification on is what exactly happens on a power outage/restore for systems. If the "managenent" system is down, how do other systems come back or even the management PC itself? Does, in the case of load sheding, does the management console control power to the UPS sockets?
Does the UPS have "smarts" to tell systems to boot up or power certian sockets?

When a system is told to reboot on power restore, what is happening on the host and/or ups to enable this? For windows servers is it just the bios setting that, if their's power boot up and the UPs is only powering on the socket? For Unix systems, is the the same?

Any insight greatly appreicated!!
Norm