Keynote
summary
Flink claims it eliminates trade-offs between volume, latency and accuracy and now runs at production scale: more than 200 contributors, an Alibaba deployment on ~1000 nodes using 5000 vcores and jobs processing millions of events/sec, King runs jobs with 30+ operators handling 30 billion events daily, and another user runs 30 Flink apps processing 10 billion events/day. The platform provides exactly-once checkpoints, event-time with watermarks, triggers, large-state support (hundreds of GB), streaming SQL, dynamic scaling and enterprise security/connectors.
AI-written summary · about
chapters (10)
- 0:00Welcome and conference overview
- 2:10Sponsors and program details
- 4:16Streaming ecosystem trends
- 13:50Flink’s role in streaming
- 20:40Flink’s future roadmap
- 27:03Operational features in Flink
- 35:07Streaming SQL and broader audiences
- 37:37Batch and streaming convergence
- 43:12Stateful stream processing evolution
- 48:07Summary and closing
related talks
40mInteractive Session - Live Code Contribution to Apache Flink
01h 07mKeynote + Conversation
16mA year in the Apache Flink Community
01h 50mKeynote
39mKeynote: CNCF Project Updates
10mKeynote
41mCloud Foundry Weekly - Live at Cloud Foundry Day NYC
19mKeynote: The Makers of Marvels: How Developers Are Rebuilding the Enterprise, One Brick at a Time