Cloud Computing and Distributed Systems Fundamentals

What Is Cloud Computing

Cloud computing delivers computing resources (servers, storage, databases, networking) on demand over the internet.

On-Premises vs Cloud:
On-Premises: buy, install, and manage your own servers
→ High upfront cost, slow to change, requires dedicated staff

Cloud: use and pay for only what you need
→ No upfront cost, elastic scaling, reduced management burden
→ Long-term cost comparison analysis still needed

Cloud Service Models

IaaS (Infrastructure as a Service):
→ Provides virtual servers, storage, and networking
→ User manages OS, middleware, and applications
→ Examples: AWS EC2, Google Compute Engine

PaaS (Platform as a Service):
→ Provides a development and deployment platform
→ User only manages application code (infrastructure is automated)
→ Examples: Google App Engine, Heroku, AWS Elastic Beanstalk

SaaS (Software as a Service):
→ Subscribe to finished software
→ Provider manages infrastructure, platform, and application
→ Examples: Gmail, Slack, Salesforce, Office 365

Key Cloud Services (AWS)

Compute:
EC2: Virtual servers (instances)
Lambda: Serverless function execution (event-driven)
ECS/EKS: Container management (Docker/Kubernetes)

Storage:
S3: Object storage (images, files, static assets)
EBS: Block storage attached to EC2 (virtual HDD)
EFS: Shared file system

Databases:
RDS: Managed relational DB (MySQL, PostgreSQL)
DynamoDB: Managed NoSQL
ElastiCache: Redis/Memcached cache

Networking:
CloudFront: CDN (global content delivery)
Route 53: DNS service
VPC: Virtual Private Cloud

Scalability

Vertical Scaling (Scale Up):
→ Upgrade server hardware (add CPU, RAM)
→ Has physical limits, single point of failure

Horizontal Scaling (Scale Out):
→ Add more servers
→ Theoretically unlimited
→ Requires a load balancer

Load Balancing:
→ Distribute traffic across multiple servers
→ Round Robin, Least Connections, IP Hash
→ Health checks automatically remove failed servers

Microservice Architecture

Monolithic:
→ All features in a single deployable unit
→ Simple but hard to scale and change

Microservices:
→ Each feature separated into an independent service
→ Each service deployed and scaled independently
→ Services communicate via APIs (REST, gRPC, message queues)

Pros: Independent deployment, technology diversity, fault isolation
Cons: Complex operations, distributed transactions, network overhead

Real world: Used by large-scale services like Netflix and Amazon

CAP Theorem

In a distributed system, you cannot guarantee all three simultaneously:

Consistency:
→ All nodes see the same data at the same time

Availability:
→ Every request receives a response (even with failed nodes)

Partition Tolerance:
→ System continues to operate despite network partitions

In practice:
→ P (Partition Tolerance) is required in distributed systems
→ CP: Consistency + Partition Tolerance (HBase, MongoDB with config)
→ AP: Availability + Partition Tolerance (Cassandra, DynamoDB)
→ Choose based on your trade-off requirements

IaaS(virtual servers) > PaaS(platform) > SaaS(finished app) — decreasing management responsibility Horizontal scaling (more servers) + load balancer = the core of cloud scalability CAP theorem: choose only 2 of Consistency, Availability, and Partition Tolerance Serverless (Lambda): deploy only code, no infrastructure management needed