Basic information
| Field | Details |
|---|---|
| English / full name | Amazon Managed Service for Apache Flink |
| Chinese | 托管实时流处理 |
| Japanese | マネージドリアルタイムストリーム処理 |
| Review priority | 考试频率 ⭐⭐⭐ |
| Often confused with | Kinesis Data Streams / Amazon Data Firehose / Glue Streaming ETL |
In one sentence
Runs stateful real-time computations over continuously arriving data from Kinesis Data Streams or Kafka/MSK.
Core summary
Hosts Java, Scala, or SQL Flink applications for windowed aggregation, live metrics, anomaly detection, and continuous transformation.
Study points
- Data Streams and MSK retain and transport streams; Flink computes over them.
- Firehose buffers and delivers data rather than acting as a general complex stream processor.
- Firehose is not a direct Flink source.
- Checkpoints and snapshots support state recovery.
Common pitfalls
- Its former name was Kinesis Data Analytics for Apache Flink.
- Do not mix up Studio notebooks, Flink applications, and legacy SQL applications.
Key memory
The stream is the road, Firehose is delivery, and Flink is the real-time processing plant.