Mastering Key-Ordered Message Processing in Apache Pulsar
Lari Hotari

In this technical deep dive, discover how to get the most out of Apache Pulsar’s Key_Shared subscription type—one of the most powerful features for scalable, key-ordered message processing. Used by thousands of organizations, Key_Shared enables parallel consumption and key-based routing at scale.

This talk examines where Key_Shared shines and where alternative patterns can deliver better performance or stronger ordering guarantees. Learn about edge cases involving negative acknowledgments, high-cardinality workloads, and DLQ handling, and explore upcoming Pulsar enhancements that make key-ordered processing even more robust.

You’ll also see practical alternatives, such as using partitioned topics with failover or exclusive subscriptions and deploying scalable, key-ordered workloads with Pulsar Reactive on Kubernetes. Real-world benchmarks and case studies highlight when Key_Shared is the right fit—and when it’s time to choose another approach.

Designed for streaming architects and engineers, this session delivers actionable insights and deployment-ready patterns for building resilient, high-performance Pulsar applications in production.

Lari Hotari
Engineering Coach, Streaming Customer Reliability Engineering, DataStax

Newsletter

Our strategies and tactics delivered right to your inbox

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.