Synchronous Kafka sends create a hard throughput ceiling.
This video explains why producer.send().get() never scales past ~200 events/sec.
See how blocking calls defeat Kafka’s RecordAccumulator batching.
Understand the real latency costs of network RTT and broker acks.
Learn the async send + callback model used at Uber-scale.
We deep dive into batching, linger.ms, compression, and acks.
Build Kafka producers that hit 50k+ events/sec on a single thread.