When I read this message I think this an absolutely terrible waste of time for a startup/small company. I want to spend my time building features not infrastructure.
This stuff is so trivially easy these days. 20 years ago it was hard to deploy a clustered message queue application, but now we've got so much open source tooling. Are you telling me you can't deploy RabbitMQ in an afternoon? Give it a day or two and you've got a monitoring stack and some swanky GitOps. Now your OpEx has been reduced 80x for hardly doing anything. As for continual maintenance, us-east-1 has gone down more times in the past year than my RabbitMQ cluster has gone down in 5. Because the tech really has improved, and it really is easy now.
Depends if you value your time more than minimum wage. There is certainly a time and a place for everything but I don’t really feel like being responsible for something that costs the $N a month managed. Sure I could do it myself but my time is worth more.
When you think about it, ideally most managed services have found some natural price for their services that helps make the above logic work. And if it does not, it might be actually overpriced. I suspect a lot of folks that say roll your own are undervaluing their time.
When we are talking about a $.50 a day message queue, no need to be in control.
If its a mission critical queue for the NYSE that has huge costs for downtime? Sure makes sense to be more in control as long as your control has a measured impact of less downtime.
Who wants to monitor and be oncall for a critical rabbitMQ service that someone only runs to save $15/month? Even if there's one outage it's already worth just paying for it.