Comparing Streaming Data Options For Speed, Simplicity & Safety
summary
Benchmarks use the OpenMessaging Benchmark on AWS i3en instances and compare Kafka and Redpanda at 50MB/s, 500MB/s and 1GB/s with TLS/SASL on. Redpanda (C++ per-core, no JVM) hits p99.99 ≈ 17ms at 500MB/s on 3 nodes; Kafka cannot complete on 3 nodes, requires 4–9 nodes for 113–389ms tail latency and shows >6s tails at 1GB/s on 6 nodes. Redpanda uses tiered S3 storage and Raft (no ZooKeeper), cutting production node count (18→3) and claiming ~$187k vs $1.2M yearly TCO (≈7× savings).
AI-written summary · about
chapters (10)
- 0:00Introduction to streaming options
- 0:24Four eras of messaging history
- 3:29Kafka's dominance and ecosystem
- 5:35Introducing Redpanda
- 8:28Benchmarking with Open Messaging
- 12:03Latency comparison results
- 14:13Cluster sizing and retention
- 17:24Cost comparison breakdown
- 20:14Redpanda deployment options
- 23:22Q&A and resources
related talks
01h 28mHigh Performance Stream Processing
38mData Pipeline with Apache Kafka
30mFive Challenges in Real-time Stream Processing and Five Solutions
39mEvaluating Streaming Framework Performance for a Large-Scale...
44mData Streaming: Which horse to bet on?
32mShould you read Kafka as a stream or in batch? Should you even care?
50mApache Kafka + Zookeeper = 3.5 Million Writes Per Second by Design
20mAnalyzing streaming data with Apache Druid
