Cloud & AWS

Virtual servers, instance types, auto-scaling groups.

• •On-demand, reserved, and spot instance pricing models
• •Auto-scaling groups adjust capacity based on demand
• •Instance types optimized for compute, memory, storage, or GPU
• •Use for: general-purpose workloads, legacy apps, full OS control

Serverless functions, cold starts, triggers, concurrency.

• •Pay only for execution time, no server management
• •Cold starts add latency on first invocation (mitigate with provisioned concurrency)
• •Triggered by API Gateway, S3 events, SQS, DynamoDB streams, and more
• •Use for: event-driven workloads, microservices, lightweight APIs

Container orchestration — Docker on AWS (ECS) and Kubernetes on AWS (EKS).

• •ECS is AWS-native; EKS runs standard Kubernetes
• •Fargate launch type removes need to manage EC2 instances
• •Trade-off: ECS is simpler, EKS offers portability and ecosystem
• •Use for: containerized microservices, CI/CD pipelines

PaaS for deploying apps without managing infra.

• •Handles provisioning, load balancing, scaling, and monitoring
• •Supports Node.js, Python, Java, .NET, Go, Docker, and more
• •Trade-off: less control than raw EC2, but much faster to set up
• •Use for: quick deployments, prototyping, small-to-medium apps

Object storage, buckets, lifecycle policies, storage classes.

• •Object storage with virtually unlimited capacity
• •Storage classes: Standard, IA, Glacier for cost optimization
• •Trade-off: eventually consistent for overwrites (strong for new puts)
• •Use for: static assets, backups, data lakes

Block storage for EC2, snapshots, volume types.

• •Persistent block storage attached to a single EC2 instance
• •Volume types: gp3 (general), io2 (high IOPS), st1 (throughput)
• •Snapshots for backups, stored incrementally in S3
• •Use for: databases, boot volumes, apps needing low-latency disk

Managed file system, shared across instances.

• •NFS-based file system mountable by multiple EC2 instances
• •Automatically scales storage up and down
• •Trade-off: higher latency than EBS, but supports shared access
• •Use for: shared config, CMS storage, container persistent volumes

Archival storage, retrieval tiers.

• •Extremely low-cost storage for rarely accessed data
• •Retrieval tiers: Expedited (1-5 min), Standard (3-5 hrs), Bulk (5-12 hrs)
• •Trade-off: very cheap storage, but retrieval has delay and cost
• •Use for: compliance archives, long-term backups, audit logs

Virtual private cloud, subnets, route tables, NAT gateways.

• •Isolated network within AWS with full control over IP ranges
• •Public subnets (internet-facing) vs private subnets (internal only)
• •NAT gateways let private subnets access the internet outbound
• •Use for: network isolation, security boundaries, multi-tier architectures

DNS service, routing policies, health checks.

• •Managed DNS with 100% SLA uptime
• •Routing policies: simple, weighted, latency-based, failover, geolocation
• •Health checks can trigger failover to backup endpoints
• •Use for: domain management, traffic routing, disaster recovery

CDN, edge locations, caching behavior.

• •Content delivery network with 400+ edge locations globally
• •Caches static and dynamic content close to users
• •Integrates with S3, ALB, API Gateway, and custom origins
• •Use for: static sites, API acceleration, video streaming

Managed API endpoints, throttling, auth integration.

• •Fully managed REST and WebSocket API service
• •Built-in throttling, request validation, and usage plans
• •Auth via Cognito, Lambda authorizers, or IAM
• •Use for: serverless APIs, microservice facades, third-party integrations

ALB vs NLB, target groups, health checks.

• •ALB: Layer 7 (HTTP) — path/host routing, sticky sessions
• •NLB: Layer 4 (TCP/UDP) — ultra-low latency, static IPs
• •Target groups route to EC2, containers, IPs, or Lambda
• •Use for: distributing traffic, high availability, blue-green deployments

Managed relational databases (PostgreSQL, MySQL), read replicas, Multi-AZ.

• •Managed PostgreSQL, MySQL, MariaDB, Oracle, SQL Server
• •Multi-AZ for high availability with automatic failover
• •Read replicas for scaling read-heavy workloads
• •Use for: traditional CRUD apps, relational data, ACID transactions

NoSQL key-value, partition keys, GSIs, capacity modes.

• •Fully managed NoSQL with single-digit millisecond latency
• •Partition key design is critical for performance and cost
• •On-demand vs provisioned capacity modes
• •Use for: high-scale apps, session stores, real-time leaderboards

Managed Redis/Memcached, caching layer.

• •In-memory caching with sub-millisecond response times
• •Redis: data structures, persistence, pub/sub. Memcached: simpler, multi-threaded
• •Trade-off: Redis is more feature-rich, Memcached is simpler to scale horizontally
• •Use for: session caching, database query caching, rate limiting

Data warehousing, columnar storage, analytics.

• •Columnar storage optimized for analytical queries (OLAP)
• •Massively parallel processing across multiple nodes
• •Redshift Spectrum queries data directly in S3 without loading
• •Use for: business intelligence, reporting dashboards, large-scale analytics