This is sort of an interim summary.

Several people have been discussing the possibility of adding a "hot backup" 
option to Pound. The idea is to have a back-end that is used only in case of 
failure of the "main" back-end(s).

Given the complexity of the proposed solutions I feel at this point that 
something like heartbeat (or similar, dedicated systems) is a much better 
solution. The soon-to-come version 2.0 of heartbeat should even allow for 
multiple server failover, so it could be a solution for people trying to 
address that situation.

This is not to mean that I would not be willing to revisit this if there is 
sufficient demand and if someone can come up with a decently simple solution, 
in line with the traditional "keep it simple" Pound approach.

One final remark: quite a few people seem to put something like

BackEnd 1.2.3.4,80,1,80

in their config files. Not only does this not bring any benefit, but it 
actually creates extra network traffic and CPU load for no good reason.

There is no situation where specifying the HA_port at the same number as the 
main back-end port can be beneficial. The HA_port was meant for those 
applications that offer an additional (supervisory) port, allowing to check 
on their health by periodically connecting there, rather than waiting for a 
connection to the main port to fail. It can also be used for taking back-ends 
off-line (and back on-line) in a controlled manner.[...]