Day 1 - Domain Review 1: Cloud Concepts + Cloud Technology and Services Essentials

📌 Key Takeaways

• The two areas we cover today (Domain 1 + Domain 3) alone make up 58% of the whole exam.
• Domain 1 is "why the cloud?" and Domain 3 is "so what do we use?"
• The axes that most often split in concepts are three: scalability vs elasticity, availability vs durability, and Region vs AZ vs edge.
• For services, memorize not the name but the condition "in what situation, this one." Questions always describe a situation.
• Treat the storage trio (S3, EBS, EFS) and the database trio (RDS, DynamoDB, Redshift) as appearing in every sitting and prepare accordingly.
• Today's goal is not reading but covering and saying it aloud (retrieval practice).

The final week has begun. This week we concentrate not on piling up new knowledge but on organizing what you have already learned so you can pull it out on the exam. Today we sweep through the two largest of the four CLF-C02 exam domains at once: Domain 1 (Cloud Concepts, 24%) and Domain 3 (Cloud Technology and Services, 34%).

Rechecking the Exam Domain Weights

Domain 3 is the largest at 34%, and together with Domain 1 it is 58% of the total. Just solidifying these two brings you much closer to the passing line (about 700/1000).

How to Use the Review Sessions

From now on it is not time for new reading but time for pulling things out. Skimming with your eyes easily creates the illusion of "I know it," but what you need in the exam room is the ability to pull it out with nothing in sight. That is why today's document has many tables. Each time you see a table, cover the right (answer) column, look only at the left and answer aloud, then mark only the items where you got stuck and go back to the original session.

This approach of recalling the answer first and then checking is called retrieval practice. Even for the same time spent, it stays in memory longer than rereading several times. Because the action the exam demands is itself "retrieval," practicing with the same action is far more effective.

Domain 1 Essentials: Cloud Concepts

The 6 Benefits of the Cloud (Be Sure to Memorize)

Why these six are "benefits" is understood right away if you think of the days before the cloud. In the past you had to decide before development, as a number, how many people would come, order equipment to match, and prepare delivery, installation, power, and cooling before you could even upload code. But that forecast is wrong anyway. Buy a lot and money is tied up in idle equipment; buy little and the service stops at exactly the moment people flood in. The cloud erased this dilemma by removing "the need to guess right in advance," and most of the other benefits derive from here.

Concept Pairs That Are Easy to Confuse

Where scores actually split in the concepts domain is the four pairs below. Because the names are similar it is easy to gloss over them, but questions aim at exactly this difference.

⚠️ Trap: It is risky to choose elasticity unconditionally from just the sentence "add servers when demand grows." If a person adds them by hand, that is scalability. For the word elasticity to be the answer, expressions like "automatically" or "adjusts automatically to demand" must appear together. Conversely, choosing elasticity for the sentence "the size can be adjusted manually" is wrong.

Cloud Computing Models / Deployment Models

With IaaS the user manages up to the OS (EC2), with PaaS the user manages only code (Elastic Beanstalk), and with SaaS the user only uses it (a finished product available immediately). If the keyword "patches the operating system directly" appears, it is IaaS.

If the three models still do not feel graspable, think of them in terms of preparing a meal. Borrowing ingredients and a kitchen and cooking yourself is IaaS; handing only the recipe (code) to a kitchen that has a chef is PaaS; buying a finished dish is SaaS. The higher up you go, the more convenient, but the range you can touch shrinks. If a question says "want maximum control," it is the lower side (IaaS); if it says "want to minimize management burden," it is the upper side (PaaS/SaaS).

AWS Global Infrastructure

• Region: A geographically separated area. Data sovereignty, latency, and pricing differ
• Availability Zone (AZ): A set of physically separated data centers within a Region. The basic unit of high availability
• Edge Location: A point for caching CloudFront content

The relationship among the three is settled in one go if you view it as a containment structure.

 [ Region (e.g., Seoul) ]
    ├── Availability Zone A   ← a set of physically separated data centers
    ├── Availability Zone B   ← spreading across these = high availability
    └── Availability Zone C
 [ Edge Locations ]  ← outside the Region, many placed close to users
                        cache content to speed up responses (CloudFront)

So the answer splits like this depending on what the question asks.

⚠️ Trap: The single expression "want to reduce latency" does not decide the answer. If it is about delivering static content (images, video, files) quickly worldwide, it is Edge Locations (CloudFront), but if you want to run the application itself close to users, the answer is choosing the Region in that area. Also, an Edge Location is only a caching point and not a high-availability means that substitutes for Availability Zones.

Domain 3 Essentials: Cloud Technology and Services

Compute

If you view compute as a single ladder of "how far will I manage?", what you must memorize shrinks sharply. From the side with the most to manage: EC2 (I manage up to the OS) → ECS/EKS (container management) → Fargate (no servers) → Lambda (code only).

Trap: People often make the mistake of choosing Lambda unconditionally on seeing the word "serverless." On AWS, serverless is not a property of Lambda alone. Fargate (containers), DynamoDB (database), and S3 (storage) also have no servers for the user to manage. If the question also says "containers," go to Fargate; if it says "NoSQL database," go to DynamoDB.

Storage

The storage trio never gets confused if you turn it into storage spaces in a house.

Suppose you build a photo-sharing service. The photos users upload pile up without limit and must be downloadable directly from the web, so you put them in S3. The EC2 instances that process photos need a disk with the OS and programs installed, so EBS is attached. If you scale the processing servers to several and all of them must see the same working folder, you attach EFS. Originals that nobody looks at after three years are moved to S3 Glacier. The key point is that within one service, the four each hold a different seat.

Trap: Reflexively choosing EBS for "multiple servers must see the same data" is wrong. A requirement for multiple EC2 instances to mount and share simultaneously is EFS. Also, an instance store is a temporary disk attached to EC2, so when it stops the data disappears. It is a wrong answer in "permanent retention" questions.

Networking

It is convenient to fix networking with a city analogy too. A VPC is a fence built around the lot I was allotted, Route 53 is the phone book that tells addresses, CloudFront is a logistics hub placed in each neighborhood, and Direct Connect is a dedicated road linking our company and AWS. Because it is a separate road rather than the public road called the internet, speed and bandwidth are stable.

ELB and Auto Scaling also often appear together here. The names look similar but they do different jobs. ELB is the bank-lobby guide who distributes incoming requests among several servers, and Auto Scaling is the branch manager who opens more counters when customers flood in.

The two are not competitors but a pair. Only when Auto Scaling adjusts the number of machines and ELB distributes requests according to that number is the elasticity of "growing and shrinking automatically with demand" complete. If a question says "distribute requests evenly," it is ELB; if it says "automatically adjust the number of instances with load," read it as Auto Scaling.

Databases

"Relational + managed" is RDS, "NoSQL + serverless" is DynamoDB, and "large-scale analytics/warehouse" is Redshift. Even making just these three mappings certain lets you get many DB questions right.

Splitting databases by situation description also gets you to the answer much faster.

Trap: If you memorize the distinction between relational and non-relational as "what is stored," it wobbles. The criterion is structure. If rows and columns are fixed and tables form relationships, it is relational (RDS, Aurora); if it is a flexible structure where the attributes held may differ per item, it is NoSQL (DynamoDB). Also, ElastiCache does not replace a database but is a cache placed in front of it. Choosing it for the purpose of "keeping the original data" is a wrong answer.

Keyword → Service Translation Table

Exam questions do not name the service but describe a situation. In the end, what you do is translate the expression into a service name. Cover the left and guess the right, then cover the right and recall the left.

List of Frequently Missed Traps

There is a reason for getting things wrong. You have to know the reason too, so you do not get it wrong next time.

1. Mixing up elasticity and scalability: Without the word "automatically," it is not elasticity.
2. Treating Edge Locations as a high-availability means: The edge is a caching point; the basic unit for failure preparedness is the Availability Zone.
3. Guessing Lambda from the single word "serverless": Fargate, DynamoDB, and S3 also have no servers to manage.
4. Choosing EBS for a file-sharing requirement: Sharing used simultaneously by multiple EC2 instances is EFS.
5. Choosing an instance store for permanent retention: It is temporary storage that disappears when stopped.
6. Putting ElastiCache in the database's place: It is a cache placed in front, not the original.
7. Memorizing IaaS/PaaS/SaaS only as service names: The exam asks "who manages up to where."
8. Guessing CloudFront from "latency" alone: For content delivery it is CloudFront; for placing the app itself closer, it is choosing a Region.

One-Line Review Cards

For covering the right side and saying it aloud.

Use these to check whether you can state each service's one-line definition and representative keywords aloud. Repeat until the service comes to mind the moment you see the keyword.

Self-Check Checklist

If you can explain it, you pass. If any item stalls you, go back to that session.

• I can state five or more of the cloud's 6 benefits without looking
• I can distinguish the difference between scalability and elasticity, and between high availability and fault tolerance, each in one sentence
• I can arrange IaaS, PaaS, and SaaS by management scope
• I can draw the containment relationship of Regions, Availability Zones, and Edge Locations
• I know what to answer to the requirement "a highly available configuration"
• I can line up EC2, Lambda, Beanstalk, and Fargate in order of management burden
• I can tell by situation when to use each of S3, EBS, and EFS
• I can describe the roles of ELB and Auto Scaling without mixing them
• I can choose among RDS, DynamoDB, Redshift, and ElastiCache by situation

Tomorrow we sort out the security and billing domains.

📖 Glossary

• Scalability: The ability itself to increase resources. It need not be automatic.
• Elasticity: The property of resources growing and shrinking automatically with demand.
• High Availability: The property of a service returning to normal quickly even if part of it fails.
• Fault Tolerance: The property of continuing to operate without stopping at all even when a failure occurs.
• Region: A geographically separated AWS infrastructure area. Data location, latency, and pricing differ.
• Availability Zone (AZ): A set of physically separated data centers within a Region. The basic unit of high availability.
• Edge Location: A point that caches content close to users. Used by CloudFront.
• IaaS, PaaS, SaaS: Three tiers of renting out only infrastructure / up to the runtime environment / up to the finished product. The higher you go, the lower the management burden.
• Serverless: An approach where the user does not prepare or manage servers directly. Lambda, Fargate, DynamoDB, and others qualify.
• Object storage: A method of storing a file whole together with its name tag. S3 is the representative.
• Block storage: A storage method used by attaching it to a server like a disk. EBS is the representative.
• ELB (Elastic Load Balancing): A load-balancing service that sends incoming requests across several servers.
• Auto Scaling: A feature that automatically increases and decreases the number of instances according to load.
• In-memory cache: A layer that keeps frequently used data in memory to speed up responses. ElastiCache is the representative.

📝 Practice Questions

Click a choice to reveal the answer and explanation.

Question 1

Which cloud computing model is most appropriate when the user wants to be directly responsible for OS patching and middleware installation?

• ASaaS
• BIaaS
• CPaaS
• DFaaS

Question 2

What is the most basic unit for building a highly available architecture within a single Region?

• AEdge locations
• BDistributing across multiple Availability Zones (AZs)
• CDistributing across multiple Regions
• DA single large instance

Question 3

Multiple EC2 instances need to mount storage simultaneously and share files. Which storage service is most appropriate?

• AAmazon EBS
• BAmazon S3 Glacier
• CAmazon EFS
• DInstance store

Question 4

Which of the following correctly describes what "trade capital expense for variable expense" means among the advantages of cloud computing?

• AYou pay a fixed monthly fee regardless of usage
• BInstead of building a data center up front, you pay only for what you use
• CAll services are provided free of charge
• DYou must prepay a full year's cost up front

Question 5

Which service is most appropriate as unlimited-scaling object storage for storing static website files and large backups?

• AAmazon RDS
• BAmazon EBS
• CAmazon S3
• DAmazon Redshift