Understanding Flink SQL changelogs is essential for anyone looking to optimize performance in streaming data pipelines. In this session, we’ll demystify how changelog modes affect everything from state size and latency to network usage and compute requirements—and why even small SQL changes can have a big impact.
You’ll learn:
- What changelog modes are and how they shape Flink SQL execution
- How operators transform data as it flows from source to sink
- The role of primary keys and upsert keys in determining changelog behavior
- Practical examples of how to choose the right mode for efficiency and accuracy
By the end of this talk, you’ll know how to interpret and optimize changelog behavior in your Flink SQL pipelines to achieve better performance and resource utilization.
Perfect for: Data engineers, stream processing developers, and anyone using Flink SQL to build or tune real-time data systems.
Bonnie Varghese is a Software Engineer at Confluent where he is part of the Flink SQL & Metastore team. He’s spent the last several years working on open-source big data systems and leveraged streaming systems like Ksql, Apache Flink, Apache Kafka, and Kafka Streams to build data pipelines. His interests lie within the broad area of systems including large-scale distributed systems and stream processing.
Suhas Satish is a Senior Engineering Manager at Confluent, leading Flink SQL, Flink runtime, and streaming AI initiatives with agents. Before Confluent, he had worked at the intersection of distributed systems and machine learning since 2013. Notable past roles include Senior Manager/tech lead in building recommendation systems at Salesforce, tech lead at Castlight Health on search & relevance, and open-source contributor to several Hadoop projects, such as Hive-on-Spark, Apache Pig, and Hue, at MapR Technologies.
Recommended resources
Watch more events.
Newsletter
Our strategies and tactics delivered right to your inbox