One of the real-world issues is production application upgrade without causing
downtime. Here the objective is that if we have a cluster of servers, how do
we make use of the cluster redundancy that is available to do application upgrade
without causing downtime.

IF you have come across nice documents on web please let me know