Wednesday, July 1, 2026

Cohort 3 | Project CloudIgnite Topics: Self-Directed CloudFormation Challenge Lab, Database Services, Billing & Support Review, Cloud Architecting & Well-Architected Framework, Certification-Prep Assessment, Spot Instances Duration: ~3 hours

---

Key Takeaways

• CloudFormation validation shortcut: the fastest way to validate a template is to create the stack — if no resources fail, the template is correct
• DynamoDB is the default choice for session state, key-value storage, and flexible-schema workloads; use Scan for non-key attributes and Query for primary key lookups
• Aurora is 5× faster than MySQL and 3× faster than PostgreSQL; Redshift is for OLAP/analytics (data warehouse)
• TAM (Technical Account Manager) and < 15-min critical response are Enterprise support only; 24/7 phone/chat starts at Business
• Global services: IAM, CloudFront, Route 53 are global; EC2 and AMIs are regional (must copy AMI to use in another region)
• CloudFront delivers static content globally with low latency (CDN); S3 stores objects; Glacier archives (not for active delivery)
• Multi-AZ deployment is the answer for availability and fault tolerance questions; ELB routes to healthy instances while Auto Scaling launches/terminates based on metrics
• User data runs at initialization (after instance creation); Spot Instances offer ~90% discount but are reclaimable on ~1-3 min notice (fault-tolerant workloads only)

---

Table of Contents

1. Self-Directed CloudFormation Challenge Lab
2. Knowledge Check — Databases
3. Knowledge Check — Billing & Support
4. Knowledge Check — Cloud Architecting / Well-Architected
5. Certification-Prep Assessment (59-Question Scenario KC)
6. Spot Instances (cost aside)
7. CLF-C02 Exam Relevance — Consolidated Map
8. Glossary
9. Checkpoint Q&A Recap
10. Action Items & Housekeeping

---

1. Self-Directed CloudFormation Challenge Lab

Goal: Build a CloudFormation template on your own that stands up a working web server (VPC → subnet → internet gateway → route table → EC2 with user-data web server), then create the stack.

Ground rules

• Do it independently. You may use an LLM (ChatGPT/Gemini) or docs for specific questions (e.g. “how do I create a route table?”) — but don't ask for the full solution.
• You may reference Monday's task1 template and copy a specific section to understand/modify — but don't copy-paste the whole thing.
• Author the template on your computer (VS Code / Notepad), then upload it to the CloudFormation console and create a stack. You don't have to use the terminal — console upload is fine.

Instructor's learning philosophy

• Type every line when learning, and try to understand each one — copy-pasting a block often feels understood but isn't.
• You're not expected to memorize everything. The real skill is: given access to information (Google, docs, an LLM), can you accomplish the task and build the template?
• Validation shortcut: the best way to know your template is correct is to upload it and create the stack. If CloudFormation shows the resources running with no failed state, everything is fine — submit and end the lab.

Environment gotchas observed

• The lab shell was Ubuntu (older kernel), with no sudo and copy-paste behavior that varied by OS — another reason to prepare the template locally and just upload it.
• Use the same template format version as the reference template if unsure.

📊 Visual: Challenge-lab workflow

The self-directed loop — author the template locally, upload it to the console, create the stack, and treat "no failed resources" as your validation that it's correct.

🎯 CLF-C02 Relevant

• Medium. Reinforces CloudFormation as declarative IaC (template → stack, success = no failed resources). The exam tests concepts, not template authoring.

---

2. Knowledge Check — Databases

📊 Visual: Which database?

Match the workload to the service — DynamoDB for key-value/flexible schema, Aurora for fast relational, Redshift for analytics, and RDS when you need automated backups and point-in-time recovery.

🎯 CLF-C02 Relevant

• High. DynamoDB (NoSQL, key-value, flexible schema), Aurora (fast managed relational), Redshift (OLAP/data warehouse), and RDS automated backups/PITR are all core Technology & Services topics.

---

3. Knowledge Check — Billing & Support

• Four support plans: Basic, Developer, Business, Enterprise.
• Basic plan: no technical support cases (account/billing only).
• Technical Account Manager (TAM): Enterprise only.
• 24/7 phone & chat: available on Business and Enterprise.
• Enterprise urgent (business-critical) response: ≤ 1 hour (fastest tier, mission-critical, is even faster).
• Support cases have 5 severity levels.
• AWS Trusted Advisor — gives recommendations to save money, improve performance, improve availability/fault tolerance, and close security gaps. (Select-3 answer: single-region risk, unused/idle resources, and how the infrastructure is performing.)
• The online tool for cost savings + availability + performance + security recommendations = Trusted Advisor.

🎯 CLF-C02 Relevant

• Very high. Support-plan tiers/features, TAM = Enterprise, response-time SLAs, and Trusted Advisor's checks are staple Billing, Pricing & Support exam questions.

---

4. Knowledge Check — Cloud Architecting / Well-Architected

• Performance Efficiency pillar areas = Selection, Review, Monitoring, Trade-offs. “Traceability” is NOT one of them.
• EC2 scaling responsibilities:
  • Load balancer → sends traffic to healthy instances.
  • Auto Scaling group → launches/terminates instances based on metrics.
  • (So “send traffic to healthy instances” is not an Auto Scaling responsibility.)
• Increase availability of a web farm → deploy across multiple Availability Zones.

  🔑 Exam heuristic: whenever a question says availability / fault tolerance, the answer almost always involves multiple AZs.

• Performance efficiency design principle → use serverless architectures.
• AWS Marketplace → find, buy, and immediately start using software solutions.
• Physical data center security is typically considered at the perimeter.
• Instance metadata is retrieved from http://169.254.169.254.

  ⚠️ Exam trap: options often keep this number but swap digits — memorize 169.254.169.254.

📊 Visual: ELB vs Auto Scaling (and multi-AZ)

Two distinct jobs — the load balancer routes traffic to healthy instances while the Auto Scaling group adds/removes instances; for availability, spread across multiple AZs.

🎯 CLF-C02 Relevant

• High. Well-Architected pillars, ELB vs Auto Scaling roles, multi-AZ for availability, serverless as an efficiency principle, and Marketplace are frequently tested.

---

5. Certification-Prep Assessment (59-Question Scenario KC)

A long, scenario-based assessment designed to mirror the real practitioner exam (worked through ~25 of 59; the rest continue tomorrow). Key takeaways:

[!TIP] Exam logistics reminder: the real CLF-C02 is 65 questions in 2 hours 10 minutes (~2 min/question). Scoring ≥ 90% on this practice set is a good readiness signal.

📊 Visual: Global vs regional services

A classic exam distinction — IAM, CloudFront, and Route 53 are global, while EC2 and AMIs are region-bound.

Concepts drilled

• Data sovereignty → prioritize data governance & legal requirements (data can't leave a region; auditors in both regions).
• Global vs regional services: Global = IAM, CloudFront, Route 53. Regional = EC2, AMI (AMIs are region-bound).
• Reduce global image-load latency → CloudFront (CDN for static content: images, video, text files).
• Store the website's images → Amazon S3 (Glacier = archive, not active delivery; EBS unsuitable).
• Most human-readable CLI output → table format.
• Check permissions without executing an action → --dry-run.
• Real-time customer notifications (no client polling) → SNS (+ Lambda to process/trigger the message).
• When does user data run? → at initialization — after the instance is already created (common trick: “creation” is wrong).
• IAM policy types: identity-based (attached to a user/role/group) and resource-based (attached to a resource).
• DynamoDB selection rationale (select 3): supports document + key-value, uses an unordered collection, and provides single-digit-millisecond latency. (It has no read replicas — that's RDS — and no BI tools.)
• Data-structure/notation mapping: curly braces {} = map/dictionary/object, brackets [] = list/array, parentheses () = tuple.
• Store customer profiles indefinitely / mobile + web, high request volume → DynamoDB (use RDS/Aurora only when it says relational / transactional / ACID).
• Track monthly session data with fast repeat reads → DynamoDB + DAX (in-memory cache, reads served without hitting the table).
• Securely connect two VPCs / private subnets → VPC Peering.
• Ingress-traffic / connectivity failure (permissions already verified) → check NACL or Security Group.

📊 Visual: Deliver vs store vs archive

The content decision — CloudFront delivers static content with low latency, S3 stores active objects, Glacier archives, and EBS is block storage for a single instance (not for this).

🎯 CLF-C02 Relevant

• Very high — this section is exam practice. Especially: global vs regional services, CloudFront/S3/Glacier use cases, SNS+Lambda notifications, user-data timing, IAM policy types, DynamoDB/DAX, VPC peering, and NACL/SG troubleshooting.

---

6. Spot Instances (cost aside)

Came up while discussing cost-effective compute:

• Spot Instances offer up to a ~90% discount, but AWS can reclaim (terminate) them on short notice (~1–3 minutes).
• Use only for fault-tolerant / non-critical / backup workloads where an interruption won't hurt the business.
• Pair with Lambda (always available) to keep costs low while covering the work Spot can't guarantee.

📊 Visual: EC2 pricing models

The cost-vs-commitment spectrum — On-Demand is flexible, Reserved is cheaper for steady workloads, and Spot is deeply discounted but reclaimable, so use it only for interruptible work.

🎯 CLF-C02 Relevant

• High. Spot vs On-Demand vs Reserved pricing and appropriate Spot use cases are common Billing & Pricing questions.

---

CLF-C02 Exam Relevance — Consolidated Map

---

Glossary

• DynamoDB — Managed NoSQL key-value/document DB; flexible schema; single-digit-ms latency.
• DAX (DynamoDB Accelerator) — In-memory cache for DynamoDB; serves hot reads without hitting the table.
• Scan vs Query — Scan reads the whole table (used for non-key attributes); Query uses the primary key.
• Aurora — AWS's high-performance managed relational DB (MySQL/PostgreSQL-compatible).
• Redshift — Managed data warehouse for OLAP/analytics.
• CloudFront — Global CDN that caches static content at edge locations to cut latency.
• VPC Peering — Private network connection between two VPCs so resources can communicate securely.
• Spot Instance — Deeply discounted EC2 capacity that AWS can reclaim on short notice.
• Trusted Advisor — Advisory service checking cost, performance, availability, and security.
• TAM (Technical Account Manager) — Dedicated support contact; Enterprise plan only.
• Identity-based policy — IAM policy attached to a user/group/role. Resource-based policy — attached to a resource.
• User data — Startup script that runs at instance initialization (after creation).
• Instance metadata — Instance info at http://169.254.169.254.
• --dry-run — CLI flag that checks permissions without performing the action.

---

Checkpoint Q&A Recap

1. Fastest way to validate your CloudFormation template? → Upload it and create the stack — no failed resources = success.
2. Find a DynamoDB item by a non-key attribute? → Scan.
3. Service for OLAP/analytics? → Redshift.
4. Which plan includes a TAM? → Enterprise only.
5. What does Trusted Advisor check? → Cost, performance, availability/fault tolerance, and security.
6. Which services are global? → IAM, CloudFront, Route 53 (EC2/AMI are regional).
7. Reduce global image latency vs. store the images? → CloudFront to deliver; S3 to store (not Glacier).
8. Real-time push notifications without polling? → SNS (often with Lambda).
9. When does user data run? → At initialization, after the instance is created.
10. Increase availability? → Deploy across multiple Availability Zones.
11. Connectivity/ingress problem after maintenance? → Check NACL / Security Group.
12. Connect two private VPCs securely? → VPC Peering.

---

Action Items & Housekeeping

• Finish the CloudFormation challenge lab — create the stack, confirm no failed resources, then submit and end.
• (Optional resume) When continuing the long assessment KC tomorrow, press “Yes” to resume — pressing “No” restarts it from the beginning (24 questions remain).
• Going forward: the course is now in exam-prep / catch-up mode — the instructor will first help everyone complete missing labs and KCs, then focus on practicing for the exam. (~120 students were on track; others get catch-up help.)
• Exam facts to remember: CLF-C02 = 65 questions / 2h10m; target ≥ 90% on practice assessments.
• Timing: class runs 7:45–10:45; KCs won't start before ~9:15; a short break offered mid-KC (after ~30 questions). Lab 316 is confirmed done.

---