It's a completely different problem because by default Prometheus does not shard anything so you're bound to a single instance, where ES and Kafka are cluster based.