WHAT IS APACHE FLINK?
3B+ data points daily streaming in from 25 million customers running real time machine learning prediction
USE CASE Streaming real-time data pipelines that need to handle complex stream or batch data event processing, analytics, and/or support event-driven applications event time window job with state and connectors for basic writes to HDFS, Pulsar, Kafka. Need Event-at-a-time/microbatch, stateful/stateless operations, and exactly once or at least once Processing
TECHNOLOGY Flink performs compute at in-memory speed at any scale Flink parses SQL using Apache Calcite, which supports standard ANSI SQL Flink runs standalone, on YARN and Kubernetes
Flink