Cloud Computing Interview Questions & Notes

Cloud computing is the on-demand delivery of computing resources like servers, storage, databases, networking, and software over the internet with pay-as-you-go pricing.

Why Cloud?

• No upfront infrastructure cost
• Scale when traffic spikes
• High availability by default
• Faster deployment

Key Concept:

You rent computing power instead of buying and maintaining physical servers.

Cloud services are offered at three levels of abstraction:

IaaS (Infrastructure as a Service)

• You rent virtual machines, storage, networks
• You manage: OS, runtime, applications
• Examples: AWS EC2, Azure VMs, Google Compute Engine

PaaS (Platform as a Service)

• You deploy applications
• Platform manages: OS, runtime, scaling
• Examples: AWS Elastic Beanstalk, Google App Engine, Heroku

SaaS (Software as a Service)

• You use complete applications
• Everything is managed for you
• Examples: Gmail, Salesforce, Office 365, Dropbox

On-Premise	Cloud
High CAPEX	Low CAPEX
Manual scaling	Auto scaling
Maintenance headache	Managed services
Limited availability	Multi-region HA
Slow provisioning	Instant provisioning

When to use on-premise:

• Strict regulatory requirements
• Legacy systems that can't migrate
• Extremely predictable workloads

Public Cloud

• Shared infrastructure
• Cost-effective
• Scalable
• Examples: AWS, Azure, GCP

Private Cloud

• Dedicated resources
• More control
• Higher cost
• Examples: VMware, OpenStack

Hybrid Cloud

• On-prem + public cloud
• Common in enterprises
• Flexibility in data placement

Multi-Cloud

• Multiple cloud providers
• Avoid vendor lock-in
• Complex networking and monitoring

Advantages:

• Cost efficiency – Pay only for what you use
• Scalability – Handle traffic spikes automatically
• Global reach – Deploy in multiple regions
• Speed – Provision resources in minutes
• Reliability – Built-in redundancy and backups
• Security – Enterprise-grade security by default

Disadvantages:

• Vendor lock-in – Hard to switch providers
• Downtime – You depend on cloud provider's uptime
• Limited control – Less control than on-premise
• Compliance – May not meet all regulatory requirements
• Cost unpredictability – Bills can spike if not monitored

Virtualization means running multiple virtual machines on one physical machine using a hypervisor.

Hypervisor Types:

Type 1 (Bare Metal)

• Runs directly on hardware
• Used in production
• Examples: VMware ESXi, Hyper-V, KVM

Type 2 (Hosted)

• Runs on host OS
• Used for testing
• Examples: VirtualBox, VMware Workstation

Why virtualization matters:

• Better hardware utilization
• Isolation
• Scalability

Virtual Machines

• Each VM has its own OS
• Heavy
• Slower boot time
• More isolation

Containers

• Share host OS kernel
• Lightweight
• Faster startup
• Better resource utilization

When to use what:

• VMs: Need different OS, strong isolation, legacy apps
• Containers: Microservices, fast deployment, resource efficiency

Scalability

Ability to increase capacity.

• Vertical: Bigger machine (scale up)
• Horizontal: More machines (scale out) – preferred

Elasticity

Ability to automatically scale based on real-time demand.

• Responds to traffic changes
• Scales up when needed, scales down when idle
• Cost optimization

High Availability

• System minimizes downtime
• Uses redundancy
• Example: Multiple instances behind a load balancer
• 99.9% uptime

Fault Tolerance

• System continues working even when components fail
• More expensive
• Used in mission-critical systems
• Example: Active-active replication across regions

High Availability	Fault Tolerance
Minimizes downtime	Zero downtime
Acceptable brief outages	No outages
Cost-effective	Expensive

What is a Virtual Machine?

A Virtual Machine (VM) is a software-based computer running on shared physical hardware using virtualization. Each VM has:

• Its own OS
• Allocated CPU, RAM, storage
• Network interface

Think of it as renting an apartment in a huge building. You don't own the building, but you control everything inside your flat.

Core VM Concepts Interviewers Expect

Instance Types

VMs come in predefined configurations optimized for different workloads:

• Compute optimized – CPU-heavy tasks
• Memory optimized – In-memory apps, caching
• Storage optimized – Databases, analytics
• General purpose – Balanced workloads

AMI / Image

A machine image contains:

• OS
• Pre-installed software
• Configuration

Launching a VM = booting from an image.

VM Lifecycle

1. Launch
2. Running
3. Stopped
4. Terminated

When to Use Virtual Machines

• Legacy applications
• Long-running services
• Need full OS control
• Custom networking or security requirements

VM Pros & Cons

Pros:

• Full control
• Predictable performance
• Mature and stable

Cons:

• Manual scaling
• Idle cost when unused
• Slower startup compared to containers/serverless

What is Auto Scaling?

Auto Scaling automatically increases or decreases the number of VM instances based on demand.

Purpose:

• Handle traffic spikes
• Maintain performance
• Reduce cost

How Auto Scaling Works

Auto Scaling is based on metrics + policies.

Common Scaling Metrics

• CPU utilization
• Memory usage
• Request count
• Network traffic

Scaling Types

Scale Out (Horizontal Scaling)

• Add more instances
• Preferred in cloud environments

Scale In

• Remove unused instances
• Saves cost

Scaling Policies

Target Tracking

"Keep CPU at 60%."

Cloud handles the math.

Step Scaling

Scale in steps based on thresholds.

Scheduled Scaling

Scale at fixed times (predictable traffic).

Benefits:

• Cost optimization
• Performance stability
• Handles unpredictable loads

What is a Load Balancer?

A load balancer distributes incoming traffic across multiple backend instances to:

• Prevent overload
• Improve availability
• Enable scaling

Layer 4 Load Balancer (Transport Layer)

Works At:

• TCP / UDP level

Characteristics:

• Routes traffic based on IP + port
• Fast
• Less intelligent

Use Cases:

• Low latency systems
• Real-time applications
• Simple traffic forwarding

Layer 7 Load Balancer (Application Layer)

Works At:

• HTTP / HTTPS level

Characteristics:

• Content-based routing
• URL-based routing
• Header-based routing
• SSL termination

Use Cases:

• Web applications
• Microservices
• Path-based routing (/api → API servers, /static → CDN)

L4 vs L7 Comparison

Feature	L4	L7
Speed	Faster	Slightly slower
Intelligence	Low	High
Routing	IP + Port	URL, headers, content
SSL Termination	✗	✓

What is Serverless?

Serverless means:

• You don't manage servers
• Code runs in response to events
• Billing is per execution time

Key Characteristics

• Event-driven
• Stateless
• Auto-scaled by cloud provider
• Short-lived execution

Common Triggers:

• HTTP requests
• File uploads
• Database changes
• Scheduled jobs

Serverless Execution Flow

1. Event occurs
2. Function is invoked
3. Code runs
4. Function shuts down

No idle cost. Finally, something efficient.

When to Use Serverless

• APIs
• Background jobs
• Event processing
• Spiky or unpredictable traffic

Cold Start Problem

A cold start is the delay when a function is invoked after inactivity.

Why it happens:

• Function needs to be initialized
• Runtime environment needs to spin up

Solutions:

• Provisioned concurrency (keep functions warm)
• Use lightweight runtimes
• Optimize initialization code

Pros:

• No server management
• Automatic scaling
• Pay per execution
• Fast deployment

Cons:

• Cold start latency
• Execution time limits
• Vendor lock-in
• Debugging complexity