Night 34 — Week 5 review + practice exam block

~2 hr warmup + 130 min timed mock (separate block). $0 AWS spend for the review and in-repo quiz. Optional: run night-34-lab-study-teardown.ps1 after the mock if pausing study for a week. Synthesizes Nights 30–33 and cross-week exam traps before Practice Exam 1.

Plain language first

What Week 5 was about

Week 4 answered “how do we make it fast and route traffic?” Week 5 answers “how do we pay for the right capacity, orchestrate the pipeline, and explain the whole system to an interviewer?”

Tonight you close the loop — one cost + orchestration decision tree, one full-stack + cost diagram, one in-repo quiz, one external 65-question timed mock, and one master teardown + portfolio checklist.

Block 1 (~20 min) — Cost + orchestration decision tree

Assignment: Draw the tree below from memory. Time yourself: 12 minutes draw, 8 minutes explain aloud using GSA examples for each leaf.

Master tree — storage cost, compute cost, orchestration

What cost problem are you solving?
│
├─ Object storage aging / unknown access pattern
│     ├─ Predictable age tiers (90d IA, 365d Glacier)
│     │     → S3 lifecycle rules on prefix (Night 30)
│     ├─ Unpredictable hot/cold per object
│     │     → S3 Intelligent-Tiering
│     └─ Finance needs spend alert at $X
│           → AWS Budgets (not Cost Explorer alone)
│
├─ Container batch runs monthly, tolerates interrupt
│     ├─ Right-size CPU/memory first (p95 utilization)
│     │     → Lower task def revision (Night 31)
│     ├─ Then purchase model
│     │     → Fargate Spot for Iceland; Compute SP for steady mixed compute
│     └─ Idle VPC tax while batch is off
│           → NAT Gateway hourly — S3/ECR VPC endpoints (Nights 8–9, 31)
│
├─ Multi-step pipeline with per-step retry and operator visibility
│     ├─ Single cron → one API call
│     │     → EventBridge schedule (Night 12)
│     ├─ Ordered steps + Catch + audit history
│     │     → Step Functions Standard (Night 32)
│     └─ Sub-5-min millions of flows, history optional
│           → Step Functions Express (exam only — not Iceland)
│
└─ Explain design to interviewer / exam scenario
      → Six tiers + Well-Architected pillar per tier (Night 33)

Cost tools map (draw second diagram)

                    Monthly bill question
                           │
           ┌───────────────┼───────────────┐
           ▼               ▼               ▼
    Cost Explorer      AWS Budgets          CUR
    "what spent?"      "alert at $X"    "line items to S3"
           │               │               │
           └───────────────┴───────────────┘
                           │
              GSA examples tonight
                           │
     ┌─────────────────────┼─────────────────────┐
     ▼                     ▼                     ▼
iceland/ lifecycle    Fargate task hours    NAT + ElastiCache idle
(Night 30)            (Night 31)            (Nights 9, 23, 31)

Checklist — labels the SAA exam expects on your diagram

Three sentences to practice out loud

1. “Right-size Fargate before buying Savings Plans; lifecycle old iceland/ months before debating Glacier Deep Archive for this month’s run.”
2. “Step Functions is the printed runbook; EventBridge cron is one line on the whiteboard — Iceland needs both history and Catch, so Standard wins.”
3. “CloudFront OAC and Cognito JWT are Security-pillar edge wins; NAT Gateway idle tax is the Cost-pillar network win — endpoints first, tear down NAT when labs pause.”

Block 2 (~40 min) — Week 5 + cross-week recap

Cost optimization ladder (Nights 30–31)

Old pipeline prefixes read quarterly?
  └─ S3 lifecycle: Standard → Standard-IA (90d) → Glacier (365d)

Current month hot output?
  └─ S3 Standard — not One Zone-IA or Glacier on upload

Unpredictable per-object access?
  └─ Intelligent-Tiering (monitoring fee)

Email when forecast > 80% of budget?
  └─ AWS Budgets — not Cost Explorer saved report alone

Hourly billing export to Athena?
  └─ CUR to S3

Fargate p95 CPU 22%, memory 35% on 2 vCPU / 4 GiB?
  └─ New task def revision — then consider Spot / Savings Plan

Monthly batch, 2-min interrupt OK?
  └─ Fargate Spot — not 24/7 On-Demand service

Flexible 1-yr commit across Lambda + Fargate + EC2?
  └─ Compute Savings Plan — not RDS RI

Orchestration ladder (Night 32)

Well-Architected + tiers (Night 33)

Cross-week traps (mock exam favorites)

Block 3 (~20 min) — In-repo Week 5 quiz

File: night-34-quiz.json — 30 scenario questions (~48 min at 96 sec each, or ~35 min if warmed up).

Topics span Week 5 (lifecycle, Fargate cost, Step Functions, Well-Architected) plus cross-week synthesis traps from Nights 1–33. Answer without night-34-quiz-answers.json first.

Score on the study site: /aws-solutions-architect-study/quiz/34/ (restart npm run dev after adding new quiz files).

Target: ≥24/30 (80%) before starting the external mock.

Block 4 (~130 min) — Practice Exam 1 (external timed mock)

Format: Real SAA-C03 — 65 questions, 130 minutes, pass bar ~72% (47/65).

Sources (pick one you already own):

• Tutorials Dojo SAA-C03 practice exam
• Stephane Maarek / Neal Davis Udemy course — Practice Test 1
• AWS Skill Builder — Exam Prep Standard / Official Practice Question Set

Rules:

1. No notes, no AWS console — one sitting.
2. Mark every question: confident / guess / skipped.
3. After submit, review every wrong answer — write the trap in one line (same format as nightly quiz notes).
4. Log score in saa-c03-exam-checklist.ts when you update the study log.

Domain mix check (SAA-C03 weights):

Target: ≥49/65 (~75%) before Night 35 weak-area drill. If below 47/65, schedule Night 35 before Practice Exam 2 — do not rush the real exam.

Block 5 (~15 min) — Portfolio checklist + teardown order

Portfolio checklist (before scheduling real exam)

Master teardown order (all study stacks)

Run when pausing labs 1+ weeks — saves NAT, Aurora, ElastiCache, idle EC2 charges.

One command (calls scripts in safe order):

.\HTML\study-lab\night-34-lab-study-teardown.ps1

bash HTML/study-lab/night-34-lab-study-teardown.sh

Pass -WhatIf (PowerShell) to print steps without running.

Cost reminders:

• NAT Gateway ~$32/mo + per-GB — step 14; biggest line item if VPC idle.
• ElastiCache cache.t4g.micro — step 6.
• Aurora Serverless v2 — step 10; scales to min ACU while cluster exists.
• Route 53 health check — step 4.

Do not tear down (production):

• globalskiatlas.com / witcoskitech.com CloudFront, WAF, OAC
• Wiki API Lambda, Cognito pool, DynamoDB visitor tables
• saa-study/gsa-wiki-cognito secret until wiki code migrates off it
• Prod lifecycle rules and Budgets that protect spend

Read-only lab artifacts (local JSON only — no AWS):

• night-33-lab-architecture-teardown — removes night-33-architecture-result.json
• night-31-lab-fargate-rightsizing-teardown — removes night-31-fargate-rightsizing-result.json

Week 5 master map

5 flashcards (write tonight)

1. Cost tree — Current iceland/2026-07/ vs 12-month-old prefix?
   (This month → Standard; lifecycle moves aged objects to IA/Glacier — Night 30.)

2. Fargate order — p95 22% CPU on 2 vCPU / 4 GiB — Savings Plan or smaller task def first?
   (Right-size task def first; then Spot / Savings Plan — Night 31.)

3. Orchestration — Iceland 2 hr, SQS on success, SNS on failure, per-step retry?
   (Step Functions Standard with runTask.sync — Night 32.)

4. Security pillar — Strongest edge control on wiki POST?
   (Cognito JWT at API Gateway / Lambda — Night 33.)

5. Teardown — Biggest idle study-VPC charge after pausing labs?
   (NAT Gateway — tear down Night 9 VPC last after dependents — Nights 8–9, 34.)

Night 35 preview

Weak-area drill — re-study top 3 missed domains from Practice Exam 1, 40 targeted questions, redo 5 missed scenarios aloud. No new deploy.

Reference links

• AWS Well-Architected Framework
• S3 lifecycle configuration
• Fargate pricing
• Step Functions Standard workflows
• SAA-C03 exam guide