Keynote: Stream analytics made real with Pravega and Apache Flink
summary
Combines Pravega, Apache Flink and Kubernetes into Dell EMC’s Streaming Data Platform and claims streams are a storage primitive that enable elastic segments, stream cuts and watermarks so jobs get repeatable historical results even as partitions change. Demos show live video ingestion with 20 ms live latency and GPU object detection at 1–2 s detection latency, DVR-style replay and reprocessing from arbitrary offsets, auto-scaling Flink parallelism, exactly‑once end‑to‑end semantics, and a Pravega Search beta (Lucene under the hood) for continuous queries.
AI-written summary · about
chapters (12)
- 0:00Introduction and time in streaming
- 3:13Stream as a new file abstraction
- 8:19Pravega streams explained
- 9:23Security camera demo
- 12:26Time-aligned video processing
- 14:21Object detection demo
- 17:36Auto-scaling with Flink
- 18:39Construction site monitoring
- 21:16Manufacturing use case
- 27:28Introducing Pravega Search
- 32:04Stream-oriented search approach
- 33:48Summary and open source
related talks
34mKeynote | Flink Ahead 2.0: The Sequel
29mA brief history of time with Apache Flink
22mCloudera's Data-in-Motion vision
36mBuilding a Self-Service Streaming Platform at Pinterest
27mArchitecting Sub-Second Analytics Pipelines
29mThe need for speed: Data streaming architectures
25mMixing data streams at scale
32mDiscover the Modern Way to Build Real-Time Analytics Solutions at Any Scale