i can explain from an linux/cluster/db point of view.
I was oracle dba as full time job for many years and there are several steps and repetition.
on oracle, versions have been
11.2 12.1 12.2
environents can be
single
multi node
and you get a mix of versions.
say you understand a single node cluster is similar to a multi node cluster, for some high level test you can have 1 node cluster
or sometimes you need a 2 or more node clusters
in oracle world, a process usually goes like this
create vm
setup disk
setup network
add additional nodes
install os
install deps
install cluster
install db sw
create dummy db
here is where automation helps, doesn't matter which you use.. shell scripts, puppet, ansible, the power is the time you free of doing already known task, so you can focus on the specific test or feature you want to use
may sound cliche, but i assure you is not
with automation, i did multiples things.
automated a 2 node cluster, and documented how to add a 3rd one.
automated a 2 node cluster, destroy one and document how to recover
automated a 2 node cluster, documented how to upgrade
etc, etc, etc
with an internet friend we did cross a vagrantfile project i did with his ansible modules and we created this:
is a end to end oracle rac cluster that is used by lots or consultants to spin nodes, we have got even ppl from oracle it self using it
on a home machine a setup can take 90 mins that you just wait, vs each install by hand can take half a day or more
so automation gives you free time to focus on what matters