Optimizing Your Iceberg Tables for Real Time Analytics (Without Melting Your Budget)

​About

Optimizing Apache Iceberg for real time analytics is not just about getting the table format right. It is about keeping tables lean, well organized, and ready to serve fresh, low latency queries while data is continuously ingested. Without the right approach, small file storms from streaming and CDC, growing delete files, and unnecessary full table rewrites can slow dashboards and operational analytics while driving costs up.

​In this webinar, we will share a practical blueprint for running fast and cost efficient Iceberg tables in production for real-time analytics across engines. We will cover usage aware optimization strategies, including targeted compaction strategies, right sized data and manifest files, effective management of equality and position deletes, and maintenance patterns that work with streaming and CDC pipelines instead of conflicting with them. We will also discuss freshness-oriented design choices such as partitioning and ordering for common filters, dynamic file sizing that adapts to ingestion rates, prompt rewrite of deletes in high traffic partitions, and metadata compaction to keep planning predictable. Finally, we will show how lightweight, specialized engines for table maintenance and optimization can replace oversized infrastructure so you achieve consistent low latency, predictable SLAs, and lower total cost without adding complexity.

Speakers
- Yingjun Wu, Founder and CEO, RisingWave
- Yuval Yogev, Co-founder and CTO, Ryft