this being my first read of the article, the following caught my attention:

"...generally I recommend log-shipping for fail-over, followed by transactional replication with queued updating subscribers. The order is reversed...