Amazon EMR Serverless
Spark/Hive without a cluster
Run Spark and Hive jobs without configuring a cluster: EMR automatically provisions and releases resources per job.
- Cost$$ Medium
- Difficulty●●●●●
- Costs money when idle?No
- Serverless
- In examsSAA
Think of it asLike hiring workers per job instead of keeping a full crew on standby.
When to use it
✓ Use it when
Spark jobs that run periodically or irregularly, when you don't want to manage a cluster.
✗ Avoid it when
You need deep cluster customization or to run many frameworks → EMR on EC2.
Common uses
- Nightly Spark ETL
- Ad-hoc analysis with Spark
- Move jobs from an EMR cluster to serverless
What does it cost?
- Pricing model: Charged by the vCPU, memory, and storage the job uses (per second).
⚠️ Cost traps
Enabling pre-initialized capacity incurs charges even while waiting for jobs.
