Section 16 - Databases

RDBMS (SQL/OLTP): RDS, Aurora
NoSQL: DynamoDB (JSON), ElastiCache, Neptune (graphs), DocumentDB (Mongo), Keyspaces (Apache Cassandra)
Object Store: S3, Glacier (backups, archives)
Data Warehouse (Analytics): Redshift (OLAP), Athena, EMR
Search: OpenSearch (JSON)
Graphs: Neptune
Ledger: Amazon Quantum Ledger Database
Time series: Amazon Timestream

Managed PostgreSQL / MySQL / Oracle / SQL Server / DB2 / MariaDB / Custom
Provisioned RDS instance size and EBS volume type and size
Auto-scaling
Supports read replicas and multi-az
Security via IAM, security groups, KMS, SSL in transit
Managed and scheduled maintenance
Automated backup with PITR
Manual snapshot for longer-term recovery
RDS Custom provides access to customize underlying instance (Oracle and SQL Server)
Useful for storing relational datasets, performing SQL queries and transactions

Compatible API for PostgreSQL and MySQL for separation of storage and compute
Data is stored in 6 replicas across 3 AZ
There are a cluster of DB instances across multiple AZ with auto-scaling of read replicas
- Clusters are custom endpoints for writer and reader DB instances
Security via IAM, security groups, KMS, SSL in transit
Managed and scheduled maintenance
Aurora Serverless: For unpredictable workloads
Aurora Global: Up to 16 DB read instances in each region, less than 1 second storage replication
Aurora Machine Learning: Perform ML using SageMaker and Comprehend on Aurora
Aurora Database Cloning: New cluster from existing one, faster than restoring a snapshot
Same use cases as RDS but less maintenance, more flexibility, performance and features

A managed Redis / Memcached, it's an in-memory data store with sub-millisecond latency
Supports Redis clustering and multi-az with read replicas (sharding)
Security via IAM, security groups, KMS and Redis Auth
Backup, snapshot and PITR features
Managed and scheduled maintenance
Need to modify app code to leverage ElastiCache
Useful for frequent reads, less writes and caching results for DB queries. Common use case is storing session data for websites

Managed NoSQL database with millisecond latency
Provisioned capacity with optional auto-scaling or on-demand capacity
Could replace ElastiCache as a key/value store
Highly available, multi-az setup by default, read and writes decoupled with transaction capability
DAX cluster for caching reads, microsecond read latency
IAM for security, authentication and authorization
Can process events using different streaming services
Has a global table feature with active-active setup
Automated backups with PITR or on-demand backups
Can export to S3 without RCU within PITR window or import from S3 without using WCU
Useful for serverless app development with documents less than 100s of KB or as a distributed serverless cache

A key/value store for objects (better for bigger objects)
Serverless, infinite scaling, max. object size is 5TB, versioning capability
Tiers: S3 Standard, S3 Infrequent Access, S3 Intelligent, S3 Glacier (lifecycle policies to move objects between tiers)
Features: Versioning, encryption, replication, MFA-Delete, Access logs, etc.
Security: IAM, bucket policies, ACL, access points, object lambda, CORS, object/vault lock
Encryption: SSE-S3, SSE-KMS, SSE-C, client-side, TLS in transit, default encryption
Batch operations: For objects using S3 Batch, listing files using S3 Inventory
Performance: Multi-part upload, S3 Transfer Acceleration, S3 Select
Automation with S3 Event Notifications, SNS, SQS, Lambda, EventBridge
Useful for static files, key/value store for big files and website hosting

DocumentDB is a fully managed AWS implementation of MongoDB (NoSQL DB) with high availability and replication across 3 AZ
MongoDB is used to store, query and index JSON data
Similar deployment concepts as Aurora
Automatically grows in increments of 10GB and scales to workloads with millions of requests per second

A fully managed graph database that's highly available across 3 AZ with up to 15 read replicas
Can build and run apps working highly connected datasets, optimizing for complex and hard queries
Store up to billions of relations and query the graph with milliseconds latency
Useful for knowledge graphs such as Wikipedia, fraud detection, recommendation engines and social networking
Streams
- Get real-time ordered sequence of every change to your graph data
- Changes are available immediately after writing
- No duplicates, there is a strict order
- Streams data is accessible in a HTTP REST API
- Useful for sending notifications for certain changes, maintaining synchronized graph data with another data store, replication across regions

A managed Apache Cassandra-compatible (open source NoSQL distributed database) database service
Scalable, highly available with tables being able to be replicated 3 times across multiple AZ
Automatically scales tables up/down based on app traffic
Less than 10ms latency at any scale and can handle thousands of requests per second
Capacity: On-demand mode or provisioned mode with auto-scaling
Encryption, backup, PITR
Useful for storing IoT devices information, timeseries data, etc.

Ledgers are books for recording financial transactions, QLDB is a fully managed service used to review history of all changes made to your app data
Highly available with replication across 3 AZ
The system is immutable, meaning no entry can be removed or modified (cryptographically verifiable)
Has 2-3x better performance than common ledger blockchain frameworks
No decentralization component compared to Amazon Managed Blockchain in accordance with financial regulation rules

A fully managed timeseries database
Automatic scaling to adjust capacity and is thousands of times faster and 10% the cost of relational databases
Features scheduled queries, multi-measure records and SQL compatibility
Tiering for storing recent data in memory and historical data in a cost-optimized storage
Built-in time series analytics functions
Encryption in transit and at rest
Useful for IoT apps, operational apps, real-time analytics, etc.