Day 4 · S7 Databases & Analytics There are 77 dump questions today. Each answer sits right under its question, so cover it and commit to a letter before you read it. The dump's key is wrong four times today, on Q31, Q356, Q527 and Q586, and Q157 and Q378 are disputed. Most of this material is Domain 3 (34%). The shared-responsibility tables count toward Domain 2. S7 · Databases & Analytics Database types Type What it is On AWS --------- Relational (SQL) Tables linked to each other, with a predefined schema, queried with SQL RDS, Aurora NoSQL Non-relational. Each is built for one data model, with a flexible schema that scales out. Types: key-value, document (JSON), graph, in-memory, search DynamoDB, DocumentDB, Neptune, ElastiCache OLTP Online transaction processing: many small reads and writes (orders, users, payments) RDS, Aurora OLAP Online analytical processing: complex queries over large historical datasets Redshift Managed database vs database on EC2 Managed: AWS handles provisioning, high availability, vertical and horizontal scaling, automated backup and restore, upgrades, OS patching, and monitoring and alerting. On EC2: you can run any database engine, but resiliency, backups, patching, high availability, fault tolerance and scaling are all your job. Amazon RDS Engines: PostgreSQL, MySQL, MariaDB, Oracle, Microsoft SQL Server, IBM Db2, and Aurora (AWS's own engine). What the managed service gives you: automated provisioning and OS patching continuous backups with Point-in-Time Restore monitoring dashboards read replicas Multi-AZ for DR maintenance windows for upgrades vertical and horizontal scaling EBS-backed storage The catch: you can't SSH into the instances. Amazon Aurora Fact Detail ------ Engine Proprietary AWS technology (not open source), compatible with MySQL and PostgreSQL only Performance About 5× MySQL on RDS and 3× PostgreSQL on RDS Storage Grows automatically in 10 GB increments, up to 256 TB Cost About 20% more than RDS, but more efficient Aurora Serverless Starts and scales automatically with actual usage, needs no capacity planning, and bills per second. For infrequent, intermittent or unpredictable workloads RDS deployment options Option Purpose How it works Exam keywords ------------ Read Replicas Scale reads Up to 15 replicas serve reads. Data is written only to the main DB "read-heavy", "improve read performance" Multi-AZ High availability A standby in one other AZ takes over if the main DB or its AZ fails. Reads and writes only go to the main DB "AZ outage", "automatic failover" Multi-Region DR and global reads Read replicas in other Regions survive a Region failure and give local reads. Cross-Region replication is billed "another Region", "disaster recovery" Shared responsibility for RDS AWS You ------ The underlying EC2 instance, with SSH disabled Security group inbound rules (ports, IPs) Automated DB engine patching In-database users and permissions Automated OS patching Creating the DB with or without public access Auditing the instance and disks and keeping them working Forcing SSL through parameter groups, and the encryption setting For DynamoDB: AWS runs the infrastructure and OS, provides the endpoints, and encrypts data at rest by default. You manage access permissions (IAM), encryption options and data classification. In-memory and NoSQL databases Service What it is Exam keywords --------- ElastiCache Managed Redis or Memcached (Valkey is also offered). An in-memory store with high performance and low latency that takes load off databases for read-heavy workloads. AWS handles OS patching, setup, monitoring, failure recovery and backups "in-memory", "cache", "open-source compatible" DynamoDB Fully managed, serverless key-value NoSQL database, replicated across 3 AZs. Millions of requests per second, trillions of rows and 100s of TB, with single-digit-millisecond latency. IAM handles security, and auto scaling is built in. Standard and Infrequent Access table classes "key-value", "serverless NoSQL", "millions of requests per second" DAX Fully managed in-memory cache for DynamoDB only. Cuts latency from milliseconds to microseconds (10×) "microseconds", "cache for DynamoDB" DynamoDB Global Tables Active-active replication across Regions: read and write in any Region with low latency "multi-Region", "active-active" DocumentDB "Aurora for MongoDB": stores, queries and indexes JSON. Replicated across 3 AZs, storage grows in 10 GB increments, handles millions of requests per second "MongoDB", "JSON documents" Neptune Fully managed graph database across 3 AZs, with up to 15 read replicas. Queries billions of relations in milliseconds "graph", "social network", "fraud detection", "recommendation engine", "knowledge graph" Timestream Fully managed, serverless time-series database. Trillions of events per day, with built-in time-series analytics "time-series", "trillions of events per day" Managed Blockchain Join public blockchain networks or run a private one, using…