GitHub user RongtongJin added a comment to the discussion: POP消费的性能
压测是否开启了enablePopBufferMerge ，开启该参数可以使得broker中Buffer Merge的策略生效，ck和ack尽量在内存中抵消。 理论上，相同consumer数量下，POP性能可能要略低于PULL，POP消费链路更加复杂，内存中无法merge还会走落盘的消息链路。Pull只是单纯的拉队列中的消息而已。但POP负载均衡是消息粒度而非队列粒度，POP消费其实可以无限扩展消费端增加消费能力。 GitHub link: https://github.com/apache/rocketmq/discussions/6360#discussioncomment-5328879 ---- This is an automatically sent email for dev@rocketmq.apache.org. To unsubscribe, please send an email to: dev-unsubscr...@rocketmq.apache.org
