i know this only tangentially related to this conversation, but we don't recommend running SpiceDB containers in ECS, because they can't do peer-to-peer dispatch in ECS without additional configuration that comes with its own problems. without horizontal dispatch you're leaving a significant amount of performance on the table by reducing your cache hit rate. we ran them using ECS at my old company for a while using CloudMap as a pseudo-service-discovery mechanism and what we found is that every time the SpiceDB pods rolled for one reason or another you'd get client errors.