Not at all, just move it to the background, no different to all the other daemonised stuff you are running on your system. If it likes to write to STDOUT, redirect this to a log file (or /dev/null if you prefer).
Much as before, no problems here either. Just use sleep at the bottom of the loop to stop it thrashing.
Your call of course.
Anacron is a slighty more advanced version of cron, may or may not do what you want. Free (I think)
Control-M is a BMC product that manages scheduling at an enterprise level, it supports job dependancies, understands limited resources and works across multiple platforms. Costs a bit.

Google for more info on these
Reentrant just means that it can be run multiple times at once without going wonky. One example:
Not rentrant:

Code:

#!/bin/sh
/usr/bin/do_some_stuff > /var/log/did_some_stuff.log
do some other things
if grep 'it worked' /var/log/did_some_stuff.log
then
  /usr/sbin/assume_we_are_good_to_go
else
  echo "Argh"
fi

Reentrant:

Code:

#!/bin/sh
/usr/bin/do_some_stuff > /var/log/did_some_stuff.$$.log
do some other things
if grep 'it worked' /var/log/did_some_stuff.$$.log
then
  /usr/sbin/assume_we_are_good_to_go
else
  echo "Argh"
fi
rm /var/log/did_some_stuff.$$.log

The first example could end up reading back the log of a different instance of itself. The second version includes the current process ID in the log file which prevents this happening.

Another, simpler, way is to just look in the process table for another copy of the script, if found exit immediatly.

(Assuming system V as opposed to bsd)

Code:

#!/bin/csh
set numprocs=`ps -ef | grep -v grep | grep $scriptname | wc -l | awk '{ print $1 }'`
if ($numprocs == 1) then
    # carry on
else
   echo "Another instance is already running, exiting..."
   exit 0
endif