Databases & Analytics

AWS CLF-C02 — Relational, NoSQL, Data Warehousing, ETL, and Purpose-Built Databases

Relational vs NoSQL Databases

Feature	Relational (SQL)	NoSQL
Data Model	Tables with rows and columns, strict schema	Key-value, document, graph, wide-column; flexible schema
Relationships	Joins across tables using foreign keys	Denormalized data, embedded documents
Scaling	Primarily vertical (scale up)	Horizontal (scale out) with ease
Query Language	SQL (Structured Query Language)	Varies by database type
Best For	Complex queries, transactions, ACID compliance	High throughput, flexible schemas, low latency

Amazon RDS (Relational Database Service)

A managed relational database service. AWS handles provisioning, patching, backups, and recovery. You cannot SSH into the underlying instance.

Supported Engines (6 total)

PostgreSQL
MySQL
MariaDB
Oracle
Microsoft SQL Server
Amazon Aurora (AWS proprietary)

Key Features

Automated backups — point-in-time restore, automated snapshots
Multi-AZ Deployment — synchronous standby replica in another AZ for disaster recovery (DR). Automatic failover. NOT used for read scaling.
Read Replicas — asynchronous replicas for read scaling. Can create up to 15 read replicas. Can be cross-AZ or cross-Region.
Storage Auto Scaling — automatically increases storage when running low
Encryption at rest using KMS, encryption in transit using SSL

Exam Tip:

Multi-AZ = Disaster Recovery (high availability). Read Replicas = Read Scaling (performance). These are two different features with different purposes!

Amazon Aurora

AWS proprietary relational database, part of RDS. Cloud-optimized with massive performance gains.

5x performance over MySQL on RDS
3x performance over PostgreSQL on RDS
Storage auto-scales in 10 GB increments, up to 128 TB
Supports up to 15 Read Replicas with sub-10ms replication lag
Instantaneous failover — high availability by design
6 copies of data across 3 AZs
20% more expensive than RDS but far more efficient
Compatible with MySQL and PostgreSQL drivers

Aurora Serverless

Automated database instantiation and auto-scaling based on actual usage
Good for infrequent, intermittent, or unpredictable workloads
No capacity planning needed — pay per second

Key Point:

Aurora is NOT open source — it is AWS proprietary. But it is MySQL/PostgreSQL compatible, meaning applications can connect using standard MySQL or PostgreSQL drivers.

Amazon DynamoDB

Fully managed, serverless NoSQL database with single-digit millisecond latency at any scale.

Key-value and document database model
Scales to massive workloads — distributed, "no limit" on throughput and storage
Serverless — no servers to provision or manage
Supports Standard and Infrequent Access (IA) table classes
Low cost and auto-scaling capabilities
Data is stored across 3 AZs by default

DynamoDB Accelerator (DAX)

Fully managed in-memory cache specifically for DynamoDB
Provides microsecond latency (10x improvement)
No code changes required — compatible with existing DynamoDB APIs
Solves the "hot key" problem (too many reads on a single partition)

DynamoDB Global Tables

Make DynamoDB tables accessible with low latency in multiple regions
Active-Active replication — read/write to any region
Two-way replication between tables across regions

Exam Tip:

DynamoDB = serverless, NoSQL, millisecond latency, key-value. If the question mentions "serverless database" or "NoSQL," think DynamoDB. DAX is the in-memory cache for DynamoDB (not ElastiCache).

Amazon Redshift

Fully managed data warehouse service for Online Analytical Processing (OLAP).

Based on PostgreSQL but NOT used for OLTP (transactional processing)
Columnar storage (instead of row-based) for analytical queries
10x better performance than other data warehouses
Scales to petabytes of data
Pay-as-you-go based on instances provisioned
Has a SQL interface for queries
Integrates with BI tools like Amazon QuickSight or Tableau

Redshift Serverless

No need to provision or manage clusters
Pay only for what you use
Good for ad-hoc analytics

Key Point:

OLTP (transactions, row operations) = RDS / Aurora. OLAP (analytics, complex queries across large datasets) = Redshift. Know the difference!

Amazon ElastiCache

Managed in-memory data store/cache service. Sub-millisecond latency for caching frequently accessed data.

Supports two engines: Redis and Memcached
Reduces load on databases for read-intensive workloads
AWS manages OS maintenance, patching, backups, failure recovery
Commonly used for session management, caching database queries, and leaderboards

Feature	Redis	Memcached
Data Persistence	Yes (backup/restore)	No
Replication	Multi-AZ with auto-failover	No replication
Data Types	Rich (strings, hashes, lists, sets, sorted sets)	Simple key-value
Use Case	Complex data, HA needed	Simple caching, multi-threaded

Exam Tip:

ElastiCache = in-memory caching for databases (Redis/Memcached). DAX = in-memory caching specifically for DynamoDB. Don't confuse the two!

Amazon Athena

Serverless interactive query service to analyze data directly in Amazon S3 using standard SQL.

No need to load data — queries data in-place in S3
Supports CSV, JSON, ORC, Avro, Parquet formats
Pricing: $5 per TB of data scanned
Built on Presto engine
Use cases: business intelligence, analytics, reporting, query VPC Flow Logs, ELB Logs, CloudTrail trails

Exam Tip:

"Serverless SQL queries on S3" = Athena. Anytime you see "analyze data in S3" or "query S3 data," think Athena.

Amazon QuickSight

Serverless, machine learning-powered business intelligence (BI) service for creating interactive dashboards.

Fast, auto-scalable, embeddable, per-session pricing
Data sources: RDS, Aurora, Athena, Redshift, S3, and more
Use cases: dashboards, visualizations, ad-hoc analysis
Pay-per-session pricing model

Amazon EMR (Elastic MapReduce)

Managed cluster platform for big data processing using Hadoop ecosystem tools.

Supports Apache Hadoop, Spark, HBase, Presto, Flink
Provisions and configures EC2 instances for you
Auto-scaling and integrated with the AWS ecosystem
Use cases: data processing, machine learning, web indexing, big data analytics

AWS Glue

Serverless extract, transform, and load (ETL) service to prepare and transform data for analytics.

Fully serverless — no infrastructure to manage
Glue Data Catalog: central metadata repository, can be used by Athena, Redshift Spectrum, and EMR to discover data
Glue crawlers scan data sources and populate the Data Catalog
Converts data between formats (e.g., CSV to Parquet)

Key Point:

Glue = Serverless ETL. Glue Data Catalog = central metadata store (like a table of contents for all your data across S3, RDS, Redshift, etc.).

Other Purpose-Built Databases

Service	Type	Key Details
Amazon Neptune	Graph Database	Fully managed graph database. Use for social networks, knowledge graphs, fraud detection. Supports billions of relationships, millisecond query latency.
Amazon DocumentDB	Document Database	MongoDB compatible. Fully managed, highly available with replication across 3 AZs. Used for content management, catalogs, user profiles.
Amazon QLDB	Ledger Database	Quantum Ledger Database. Records an immutable, cryptographically verifiable journal of changes. Used for financial transactions, supply chain, registration systems. No decentralization (unlike blockchain) — central trusted authority.
Amazon Managed Blockchain	Blockchain	Join public blockchain networks or create/manage decentralized private networks. Supports Hyperledger Fabric and Ethereum.
Amazon Timestream	Time-Series Database	Fully managed, serverless. For IoT and operational applications with time-series data. 1000x faster and 1/10th cost vs relational DBs for time-series workloads.
AWS DMS	Migration Service	Database Migration Service. Migrate databases to AWS. Supports homogeneous (same engine) and heterogeneous (different engines via Schema Conversion Tool) migrations. Source database remains available during migration.

Exam Tip:

QLDB = immutable ledger, centralized (no decentralization). Managed Blockchain = decentralized, multiple parties, no central authority. Neptune = graph (social networks, recommendations). DocumentDB = MongoDB. Timestream = IoT time-series.

Database Selection Cheat Sheet

Need	Service
Relational database (managed)	Amazon RDS / Aurora
NoSQL key-value, serverless	Amazon DynamoDB
In-memory cache for DB	Amazon ElastiCache
In-memory cache for DynamoDB	DynamoDB DAX
Data warehouse (OLAP)	Amazon Redshift
SQL queries on S3	Amazon Athena
BI dashboards	Amazon QuickSight
Big data / Hadoop	Amazon EMR
ETL & data catalog	AWS Glue
Graph database	Amazon Neptune
MongoDB compatible	Amazon DocumentDB
Immutable ledger	Amazon QLDB
Decentralized blockchain	Managed Blockchain
IoT time-series	Amazon Timestream
Database migration	AWS DMS