Amazon EMR
- What it is
- Managed big data platform running open-source frameworks: Apache Spark, Apache Hadoop, Apache Hive, Apache HBase, Apache Presto, Flink.
- AWS provisions and manages the cluster; you keep full control of frameworks, versions, and tuning.
- Deployment options
- EMR on EC2 — classic clusters. EMR Serverless — no cluster sizing. EMR on EKS — run Spark on your Kubernetes cluster.
- Cost lever
- EMR clusters are the textbook use of Spot Instances for task nodes: interruption-tolerant work at up to 90% off.
- Not to be confused with
- AWS Glue — serverless, ETL-focused, minimal configuration. Choose EMR when you need specific framework versions, custom libraries, long-running clusters, or fine tuning.
- Amazon Redshift — a data warehouse for SQL analytics on structured data. EMR processes raw data at scale in any format.
- Exam signal
- “Hadoop”, “Spark”, “petabyte-scale processing with open-source frameworks” → EMR.