What's new in 1.9.0 blink planner
summary
Flink 1.9 adds a pluggable Blink planner that unifies batch and streaming under one API, swaps to a new binary row layout and type system, and implements push-based batch operators plus hash/sort algorithms with memory-fallback. It adds bundle memory aggregation to buffer records and cut state I/O, yielding observed 2x–10x throughput gains, plus local combiner and distinct-count optimizations, lookup-based dimension joins, and a dedicated top-N rank operator for efficient streaming ranking.
AI-written summary · about
chapters (15)
- 0:00Intro
- 0:14Agenda
- 1:53A Step Closer
- 3:50Backward Compatibility
- 4:39Major Blink Planner Features
- 5:10New Type System
- 8:12Binary Formats (Binary Row) • Deeply integrated with memory segments
- 11:55Distinct Aggregation Skew Handling
- 13:42Bundle Processing
- 18:23Processing Time Dimension Join
- 25:14Streaming De-duplication
- 27:18Batch Improvements
- 28:58Hash-based Algorithms
- 30:40More Functionality Coverage
- 31:21Summary & Futures
related talks
44mNews from Flink’s engine room: “Full steam ahead”
30mFlink + Pulsar: The Path To Unified Batch and Streaming
23mBetter, Faster, Stronger Streaming: Your First Dive into Flink SQL
40mThe Seven Deadly Sins of Streaming
43mUnified stream and batch processing with Apache Flink’s relational APIs
32mKeynote: Apache Flink - Completing Cloudera’s End to End Streaming Platform
04h 15mTopical Edition on Databases - Track 2
24mFlinkNDB : Skyrocketing Stateful Capabilities of Apache Flink