# STUDY_PLAN — 15 Days, Nov 4–18

## Daily repetition stack (every day, in order)

1. Say scoring words aloud (00_START_HERE).
2. Replay yesterday's audio once at 1.25x.
3. Recite intro spine from memory (CHEAT_SHEET §6).
4. One STAR story aloud with its metric.
5. New day's Focus + Listen + Lab below.
6. Done-check before bed.

## Days 1–15

### Day 1 — DSQL architecture
- Focus: disaggregated compute/storage, journal durability, HLC/TimeSync.
- Listen: E1. Lab: create single-Region cluster, pgbench smoke.
- Say-aloud: "DSQL separates compute, storage, and transaction coordination; durability comes from the journal."
- Desk: sketch 3-layer diagram. Done-check: can name all 3 layers + clock role.

### Day 2 — DSQL transactions
- Focus: MVCC, snapshot isolation, optimistic concurrency, commit-time adjudication.
- Listen: E2. Lab: force a write-write conflict, implement retry.
- Say-aloud: "Conflicts are adjudicated at commit; loser retries."
- Desk: write conflict-retry pseudocode. Done-check: explain dirty-read prevention.

### Day 3 — Multi-Region DSQL
- Focus: 2 active Regions + witness quorum, zero-data-loss failover.
- Listen: E3. Lab: multi-Region cluster, failover drill.
- Say-aloud: "Two active plus witness gives quorum without losing writes."
- Desk: draw quorum loss scenarios. Done-check: state RPO=0 claim precisely.

### Day 4 — DSQL limits + compat
- Focus: quotas (10 MiB txn, 3000 rows/txn, 10k conns, 10 TiB default/256 TiB max, 24 indexes/table, 10 schemas), PG gaps.
- Listen: E4. Lab: boundary tests. Say-aloud: recite 6 caps from memory.
- Desk: limits flashcard. Done-check: all caps without notes.

### Day 5 — Aurora Limitless
- Focus: router + shard groups, sharded vs standard tables, ACU, I/O-Optimized only.
- Listen: E5. Lab: shard-key design exercise.
- Say-aloud: "Shard by the join key you filter on."
- Desk: convert one schema to sharded. Done-check: name 3 Limitless constraints.

### Day 6 — Aurora Global DB + RDS HA
- Focus: Global DB (1 writer + ≤10 read regions, <1s lag) vs DSQL multi-active; Multi-AZ (HA) vs read replicas (scale, async).
- Listen: E6. Lab: lag reading + failover sim.
- Say-aloud: "Multi-AZ is failover, replicas are scale."
- Desk: Global-vs-DSQL table. Done-check: explain single-writer trap.

### Day 7 — DynamoDB + DAX
- Focus: Global Tables multi-active KV, DAX microsecond cache, write-bypass staleness.
- Listen: E7. Lab: DAX staleness demo.
- Say-aloud: "DAX bypass on global-table writes goes stale until TTL."
- Desk: DAX decision checklist. Done-check: when NOT to add DAX.

### Day 8 — ElastiCache
- Focus: Valkey (Redis successor), Serverless, ephemeral-cache design.
- Listen: E8. Lab: cache-aside + eviction drill.
- Say-aloud: "Cache is ephemeral; always design the miss path."
- Desk: eviction-policy picks. Done-check: Serverless vs self-managed call.

### Day 9 — DocumentDB / Keyspaces / Neptune
- Focus: doc/wide-column/graph picks + compat traps.
- Listen: E9. Lab: compat checklist per engine.
- Say-aloud: "Compat is not 100% — check APIs and indexes."
- Desk: engine-picker table. Done-check: one-sentence pick per engine.

### Day 10 — Analytics: Redshift vs Athena vs S3
- Focus: Redshift curated warehouse, Athena ad-hoc on S3, zero-ETL Aurora→Redshift.
- Listen: E10. Lab: Athena partition-pruning cost demo.
- Say-aloud: "Athena bills per scan — partition or pay."
- Desk: cost-trap list. Done-check: zero-ETL limits.

### Day 11 — DMS + SCT + replication theory
- Focus: SCT assessment, DMS full+CDC logical replication, physical vs logical.
- Listen: E11. Lab: DMS task with PK/logging prereqs.
- Say-aloud: "DMS is logical — needs keys and source logging."
- Desk: physical-vs-logical table. Done-check: 3 DMS prereqs.
- Enablement rep (posting core): whiteboard a 10-min Oracle→DSQL migration plan for a strategic customer — SCT assessment → compat-gap triage → DMS full+CDC → cutover + rollback. Close with one best-practice rule you'd publish and one product-gap feedback you'd file to Eng.

### Day 12 — S3 durability + 403 triage
- Focus: versioning, lifecycle, CRR/SRR; 403 order: Deny > IAM > bucket policy > KMS > SCP/conditions.
- Listen: E12. Lab: 403 repro.
- Say-aloud: recite 403 order from memory.
- Desk: triage runbook. Done-check: replication+versioning rule.

### Day 13 — VPC + observability
- Focus: private subnets ≥2 AZs, SG least-privilege, endpoints/Proxy; CloudWatch + PI + Enhanced Monitoring.
- Listen: E13. Lab: misconfig hunt + PI top-SQL read.
- Say-aloud: "SGs are stateful; routes and NACLs still blackhole."
- Desk: VPC diagram. Done-check: PI vs EM split.

### Day 14 — Oracle→AWS + STAR bank
- Focus: translation pairs + 6-story STAR reps + intro spine.
- Listen: E14. Lab: timed STAR reps.
- Say-aloud: full 90-sec intro. Done-check: every story has a metric.

### Day 15 — Mock loop
- Focus: customer migration whiteboard + SQL perf + LP + Bar Raiser simulation.
- Listen: E15. Lab: full timed mock, gap list.
- Rounds: (1) whiteboard Oracle→DSQL migration for a strategic customer, (2) slow-query triage on PG/DSQL with EXPLAIN, (3) LP behavioral with metrics, (4) Bar Raiser depth probe. Narrate as if the interviewer is the customer.
- Say-aloud: weakest answer 3x. Done-check: gap list ≤5 items with owners.

## BASE-TO-BUILD

| Base (23 yr Oracle) | Build (Principal DSQL, AWS) |
|------|------|
| RAC / Data Guard / GoldenGate / RMAN / AWR-ASH | Aurora Multi-AZ-Limitless-DSQL / Global DB / DMS CDC / snapshots+PITR / PI+EM+CloudWatch |
| Single-engine depth | Breadth across 11 AWS engines + DSQL depth |
| DBA-run ops | Serverless, no-OS-access managed ops |
| Enterprise customer escalations + product feedback to Eng (Oracle Principal TSE) | DSQL-specific best practices + scaled enablement via field teams |
| Runbook fixes | Fixes that stay fixed + hire-and-develop others |