Thousands a day? Really? Even if it were hundreds of thousands a day it would make more sense to use a managed Pub Sub service and save yourself the hassle (assuming modest throughput).
Yeah, I'd do the opposite of what they ended up doing. Start with Postgres, which will handle their thousands-per-day no sweat. If they scaled to > 100 millions/day, then start investigating a dedicated message bus / queue system if an optimized PG solution starts to hit its limits.
Yeah, it seems like a more natural evolution into specialization and scale rather than a step "backwards" to PG which I suspect will be the subject of a future blogpost about replacing their pq queue with another solution...
