The concern wasn't significant revenue, it was really just to cover the cost of the more robust infrastructure without a single point of failure.

The initial product was just an nginx config for me and Austin's shortlinks and domain forwards. And that actually ran just fine for months with no downtime.

Problem is then you get customers and you want to promise them no down time, so you have to allow for failovers and high availability, and that adds complexity that really isn't needed... until it is. And in the almost 3 years, we had 1 database fail over to a slave that got promoted to master, and actually stayed that way for months before I manually failed it back to the bigger server.