Analytics Services — Redshift, Athena, EMR, Kinesis, QuickSight

The 5 services

Service	Type	Mental model	Language
Redshift	Data warehouse	Big SQL analytics DB (petabyte) — LOAD data first	SQL
Athena	Serverless query on S3	SQL over files WHERE THEY SIT — no loading	SQL
EMR	Big data processing	Managed Hadoop/Spark clusters	MapReduce/Spark
Kinesis	Real-time streaming	Managed Kafka — live event pipeline	SDK
QuickSight	BI dashboards	AWS Tableau/Looker — visualize	Drag-drop

The key distinction: Athena vs Redshift

Redshift = data LOADED into warehouse → then queried (copy, optimized for analytics)
Athena   = data STAYS in S3 → SQL runs directly over the files (serverless, pay/TB scanned)

• Athena ≠ warehouse — nothing stored, nothing loaded
• Athena = query engine reading files in place
• “query data directly in S3 without loading” → Athena

Others

• RDS (transactional, current state) vs Redshift (analytical, history)
• EMR: batch processing, can use Spot (90% off)
• Kinesis components: Data Streams (raw), Firehose (→S3/Redshift), Data Analytics (SQL on stream)
• Kinesis vs SQS: streaming (multi-consumer, replay) vs queue (one consumer, jobs)

The pipeline

Sources (apps, IoT, logs)
  ↓
Kinesis (real-time) ──→ Firehose
  ↓                        ↓
S3 (data lake)      Redshift (warehouse)
  ↓
Athena (query S3) ──→ QuickSight (dashboards)
EMR (heavy Spark processing)

Exam shortcuts

Need	Answer
Big SQL analytics warehouse	Redshift
Query S3 files directly (serverless)	Athena
Real-time streaming	Kinesis
Hadoop/Spark batch	EMR
BI dashboards	QuickSight
Petabyte reporting	Redshift

Related

• [[aws/index]]
• [[aws/storage-types/index]]