Chapter Three
Databases
Eight databases spanning relational, document, key-value, in-memory, analytical, graph, and time-series models. AWS gives you a purpose-built engine for each shape of data — and the cost of picking wrong is a rewrite.
Core Terminology
A few terms recur across this chapter. Defining them here keeps the service sections focused.
OLTP vs OLAP
Transactional databases (OLTP) serve many small reads and writes; analytical databases (OLAP) scan huge ranges for aggregates. Different engines, different shapes.
Read Replica
A read-only copy that offloads read traffic from the primary. Asynchronous, so it can lag slightly behind.
Multi-AZ
A synchronous standby in another Availability Zone that takes over automatically on failure — durability and failover, not read scaling.
Partition Key
The attribute DynamoDB hashes to distribute data. Choosing it poorly creates hot partitions that throttle throughput.
Provisioned vs On-Demand
Two capacity modes: reserve throughput for predictable load and lower cost, or pay per request for spiky, unpredictable load.
Services in This Chapter
Service 16
Amazon RDS
Managed relational databases — PostgreSQL, MySQL, MariaDB, Oracle, SQL Server. AWS handles patching, backups, and failover; you keep the engine you know.
Service 17
Amazon Aurora
AWS's cloud-native MySQL- and PostgreSQL-compatible engine. Decoupled storage, faster failover, and read scaling beyond what stock RDS offers.
Service 18
Amazon DynamoDB
Serverless key-value and document database with single-digit-millisecond latency at any scale — provided you model your access patterns up front.
Service 19
Amazon ElastiCache
Managed Redis and Memcached. The in-memory cache and ephemeral store that takes read load off your primary database.
Service 20
Amazon Redshift
Petabyte-scale data warehouse. Columnar, massively parallel analytics over structured data — the OLAP counterpart to your OLTP databases.
Service 21
Amazon DocumentDB
MongoDB-compatible managed document database. The choice when you have Mongo workloads and want AWS to operate them.
Service 22
Amazon Neptune
Managed graph database for highly connected data — social graphs, fraud rings, recommendation networks — with Gremlin and SPARQL support.
Service 23
Amazon Timestream
Purpose-built time-series database for IoT and operational metrics, with tiered storage that ages old data to cheap storage automatically.