Cloud computing is on-demand delivery of IT resources and applications via the Internet with pay-as-you-go pricing

Foundational services

meta-data vs user-data vpc vs subnet

regions vs Availability Zones (AZ)

s3 vs ebs vs instance storage

• s3
  • encrypt by SDK
  • versioning->rollback
  • Lifecycle management->ex. log file, archives
• s3-IA
  • Infrequent Access
  • extra cost for retrieval
  • cheaper
• Glacier
  • 3-5 hrs/retrieve
  • cheapest
• ebs
  • SSD
    • IOPS good
  • HHD
    • throughput good
  • create snapshot and store in s3
  • loose coupling and made app stateless

CloudTrail

• records AWS API calls for accounts

RDS vs DynamoDB vs ElastiCache

• RDS
  • synchronously replicated cross AZ
  • snapshots in s3
  • encryption

auto scaling

• elb, cloudwatch, auto scaling
• what->where->when
  • launch config->security group & AMI...etc->cloudwatch alerm/sceduale event
  • Minimum/Desired/Maximum size
• Application Load Balancer->new service
  • route traffic to a target group
• cloudwatch
  • monitoring service for AWS cloud resources
  • alarms->scale in/out
  • custom metrics

Everything fails, all the time

Cloud Architecture Principle

• Design for failure
• Embrace Elasticity & Automate
• Loose coupling sets you free
  • 3 tiers, web -> app -> db
  • pipeline mode: a->b->c by queues
• Become stateless
  • ex. DynamoDB
• Think Parallel
  • 4 ec2 1 hrs === 1 ec2 4 hrs -> same price but lesser time cost
• Leverage different storage options
  • Offload log file to s3
• Build security into every layer
  • privelege -> ex. security group
  • IAM