Having jobs scheduled by cron does not seem as efficient to me as the way I have seen other monitoring tools poll. What I would like to see is some sort of daemon that would call the discovery, polling, alerting and other jobs in an organized way. By distributing this load it would be more efficient.
This reminds me of re-spinning my daemon design
I had somewhere a version that implemented network communication and was distributed using a built-in queue system (better than the memcached approach) that feels like any other AMQP