Modern enterprises need data platforms that can handle both real-time ingestion and analytical storage without complex integration layers. In this session, we introduce Ursa, a high-performance streaming engine built on Apache Pulsar, designed to unify streaming and lakehouse architectures.
You’ll learn how Ursa bridges Pulsar’s low-latency event streams with modern lakehouse systems, enabling seamless, efficient writes from topics directly into analytical storage — transforming streaming data into queryable tables in real time.
Key highlights:
- Inside Ursa’s architecture: Combining Pulsar’s scalability with the durability and structure of lakehouse storage.
- Two integration modes:Managed Tables — lightweight, internally managed storage for quick access.
- External Tables — writes data to external lakehouses with open catalog registration. Support for Databricks Unity Catalog, Snowflake’s Open Catalog, and Open S3Table format.Handling schema evolution, efficient file writing (e.g., Parquet optimization), and table registration for seamless downstream analytics.Real-world insights into building cloud-native, unified lakehouse pipelines powered by Ursa and Pulsar. Whether you’re architecting real-time pipelines, modernizing your lakehouse, or looking to simplify your streaming-to-analytics stack, this talk provides the blueprint for achieving a truly unified data platform.
