5.2 Storage and Database Deep Dive

Quick Answer: This deep dive covers the advanced data features the exam revisits: S3 versioning, lifecycle policies, Object Lock, and replication; EBS snapshots and encryption; RDS Multi-AZ vs. Read Replicas; and DynamoDB Streams, Global Tables, and DAX. The single most-tested pair is Multi-AZ (high availability) vs. Read Replicas (read scaling).

Amazon S3 Advanced Features

S3 Versioning

Versioning keeps multiple variants of an object in a bucket. Each version gets a unique ID, a delete adds a recoverable delete marker, and you can restore prior versions — protecting against accidental overwrite and deletion.

S3 Lifecycle Policies

Lifecycle policies automate transitions and expirations. A typical policy:

1. Day 0 — store in S3 Standard
2. After 30 days — transition to Standard-IA
3. After 90 days — transition to Glacier Flexible Retrieval
4. After 365 days — delete the object

S3 Object Lock

Object Lock enforces WORM (Write Once Read Many): objects cannot be deleted or overwritten for a retention period — used for regulatory compliance.

S3 Replication

Uses include compliance, disaster recovery, and lower-latency access for distant users.

S3 Transfer Acceleration

Uses CloudFront edge locations so clients upload to the nearest edge, and AWS routes the data to S3 over its optimized backbone — speeding long-distance uploads.

---

Amazon EBS Advanced Features

EBS Snapshots

• Point-in-time backups stored in S3, and incremental (only changed blocks after the first)
• Can create new volumes in any AZ or Region
• Snapshot Archive moves rarely used snapshots to a far cheaper tier
• Recycle Bin guards against accidental snapshot deletion with retention rules

EBS Encryption

Encrypts data at rest, in transit between instance and volume, and all snapshots, using AWS KMS keys with minimal latency impact. To encrypt an existing unencrypted volume, you create an encrypted snapshot and restore from it.

---

Amazon RDS Advanced Features

Multi-AZ Deployments

Read Replicas

On the Exam: Multi-AZ = high availability (automatic failover). Read Replica = read performance scaling. They solve different problems; do not swap them. A scenario about surviving an AZ outage wants Multi-AZ; one about offloading heavy read traffic wants Read Replicas.

RDS Backups

Automated backups (1–35-day retention) enable point-in-time recovery, and manual snapshots persist until you delete them.

---

Amazon DynamoDB Advanced Features

DynamoDB Streams

Captures a time-ordered log of item changes and can trigger Lambda for event-driven processing.

DynamoDB Global Tables

Multi-Region, multi-active replication — every replica accepts reads and writes with automatic conflict resolution, ideal for globally distributed apps.

DynamoDB Accelerator (DAX)

An in-memory cache delivering up to 10x read improvement, taking responses from milliseconds to microseconds.

Capacity Modes

---

AWS Backup

AWS Backup centralizes and automates backups across many services (EBS, RDS, DynamoDB, EFS, S3, and more) with backup plans, cross-Region and cross-account copies, and policy enforcement through AWS Organizations.

On the Exam: When a scenario asks for one centralized, automated backup solution across multiple services, the answer is AWS Backup — not per-service snapshots managed separately.

---

Durability, Availability, and Recovery

These features exist to deliver two often-confused properties. Durability is the probability your data is not lost — S3's eleven nines and Aurora's six copies across three AZs are durability features. Availability is the probability you can reach your data right now — Multi-AZ failover, Read Replicas, and replication address availability. A design can be highly durable but temporarily unavailable, or vice versa, so read the scenario carefully for which property is at risk.

A few high-value rules tie it together. To survive an entire AZ failure, use Multi-AZ (RDS) or spread resources across AZs. To survive an entire Region failure, you need cross-Region tooling: S3 Cross-Region Replication, DynamoDB Global Tables, Aurora Global Database, or cross-Region backups. To scale reads, add Read Replicas or a cache (ElastiCache/DAX). And to recover from accidental deletion or corruption, rely on versioning, Object Lock, snapshots, and AWS Backup retention. Mapping the scenario to the right property — durability, availability, scaling, or recovery — points directly to the intended service combination.

---

Cost-Aware Storage Choices

Finally, several of these features double as cost levers. Lifecycle policies move cold objects to cheaper S3 tiers automatically; EBS Snapshot Archive cuts the cost of rarely used backups; and choosing DynamoDB on-demand versus provisioned capacity can swing cost dramatically depending on traffic predictability. The exam often blends "most cost-effective" with a storage or database scenario, so remember that the cheapest correct design usually combines the right tier with the right replication or backup posture — not the most expensive always-on option.