kafka or rabbitmq when do you use which one

hey folks im runnin into a common crossroads and wanted to get some community input we're buildin out a new data pipeline that needs to handle high throughput and provide durable message storage both kafka and rabbitmq seem like strong contenders but they have different core philosophies kafka feels more like a distributed log great for replayability and stream processing rabbitmq on the other hand is a more traditional message broker focusing on smart routing and guaranteed delivery with features like acknowledgments and deadletter queues for a system that needs both high throughput and reliable delivery where do you draw the line when does kafkas logcentric approach win over rabbitmqs brokercentric features or vice versa what are the key tradeoffs you consider