Data Store Management
This domain tests your ability to select, configure, and manage the correct AWS storage and database service for a given analytical or operational workload. It comprises 26% of the DEA-C01 exam, making it the single largest domain. Expect questions that require you to distinguish between services based on latency requirements, query patterns, cost optimization, access control, and scalability — not just recall service names.
Amazon S3
Amazon S3 is the foundational object store for data lakes and analytics pipelines on AWS.
Storage Classes
| Storage Class | Min Duration | Retrieval Time | Use Case |
|---|---|---|---|
| S3 Standard | None | Milliseconds | Frequently accessed data |
| S3 Standard-IA | 30 days | Milliseconds | Infrequent access, rapid retrieval |
| S3 One Zone-IA | 30 days | Milliseconds | Infrequent, non-critical, single AZ |
| S3 Glacier Instant Retrieval | 90 days | Milliseconds | Archives accessed once per quarter |
| S3 Glacier Flexible Retrieval | 90 days | Minutes–hours | Archives, flexible restore |
| S3 Glacier Deep Archive | 180 days | Up to 12 hours | Long-term compliance archives |
| S3 Intelligent-Tiering | None | Milliseconds–hours | Unknown or changing access patterns |
Intelligent-Tiering automatically moves objects between tiers: Frequent Access → Infrequent Access (after 30 days) → Archive Instant Access (after 90 days) → Archive Access (after 90–180 days, optional) → Deep Archive Access (after 180+ days, optional). There is a small monthly monitoring fee per object; it is not cost-effective for objects smaller than 128 KB.
Lifecycle policies transition or expire objects on a schedule. Transitions must follow the waterfall order — you cannot move from Standard directly to Standard-IA and skip to Deep Archive in a single rule without passing through intermediate classes (though you can skip classes in rules). Minimum 30-day stay in Standard before transitioning to Standard-IA or One Zone-IA.
Versioning must be enabled before MFA Delete can be configured. Versioning cannot be fully disabled once enabled — only suspended. Event notifications (S3 → SNS, SQS, Lambda, or EventBridge) trigger on object PUT, DELETE, etc.
Common Traps
- Selecting Glacier Flexible Retrieval when millisecond restore is required — use Glacier Instant Retrieval instead.
- Applying Intelligent-Tiering to tiny objects (< 128 KB) — monitoring fees exceed storage savings.
- Assuming lifecycle transitions are instant — they run once daily.
- Confusing replication (cross-region or same-region) with versioning — replication requires versioning but they are separate features.
Worked Scenario
Requirement: A compliance team stores audit logs that must be retained for 7 years, accessed only during audits (roughly once per year), with retrieval within 12 hours acceptable.
Reasoning: 180-day minimum duration + 12-hour retrieval → S3 Glacier Deep Archive. Glacier Flexible Retrieval also fits on retrieval time but Deep Archive is cheaper for truly long-term archival.
Amazon Redshift
Redshift is AWS's petabyte-scale columnar MPP (massively parallel processing) data warehouse optimized for OLAP workloads.
Distribution Styles
| Style | When to Use |
|---|---|
| KEY | Join columns — co-locates matching rows on the same node |
| ALL | Small dimension tables — full copy on every node |
| EVEN | No clear join key, large tables needing even distribution |
| AUTO | Default; Redshift chooses KEY or ALL based on table size |
Sort Keys
- Compound sort key: columns listed in order; most effective for range queries and ORDER BY on leading columns.
- Interleaved sort key: equal weight to each column; good for multiple filter columns but has higher maintenance cost (VACUUM takes longer).
- Sort key + zone maps: Redshift stores min/max per 1 MB block; queries skip blocks outside the range (zone map pruning).
Compression and Encoding
Redshift uses columnar compression (encoding). Run ANALYZE COMPRESSION to get recommendations. ENCODE AUTO (default for new tables) lets Redshift manage encodings automatically.
Workload Management (WLM)
- Automatic WLM (default): Redshift dynamically assigns memory and concurrency.
- Manual WLM: define up to 8 queues with memory %, concurrency, and query group/user group routing rules.
- Concurrency Scaling: automatically adds transient clusters for read bursts; enabled per queue.
- Query monitoring rules (QMR): abort, log, or hop queries that exceed hop thresholds (scan rows, execution time, etc.).
Common Traps
- Choosing interleaved sort key for time-series data — compound sort key on timestamp is far more efficient.
- Using EVEN distribution when two large tables are frequently joined on the same column — causes data movement (redistribution); use KEY distribution on the join column.
- Assuming Redshift is OLTP — it is not designed for high-concurrency single-row lookups.
Unlock with a paid plan
12 more sections (about 2,853 words) continue below. Every paid plan unlocks all DEA-C01 study guides, the full question bank and mock exams, from $4.97/month.