Sure. We could also not use a container registry. Or resource management. Or liveness/readiness probes. For the database we could use RDS and take the latency/$$$ overhead. For file storage set up a separate EBS volume or whatever. But... why?