AWS Global Infrastructure Services

Summary: https://www.udemy.com/course/aws-certified-cloud-practitioner-new/learn/lecture/20260650

Global Applications

Ref: https://www.udemy.com/course/aws-certified-cloud-practitioner-new/learn/lecture/20056110 and https://www.udemy.com/course/aws-certified-cloud-practitioner-new/learn/lecture/29102360

🔧 Global application = app that is deployed in multiple geographic locations
- e.g. deploying an app in multiple AWS regions and/or edge locations
- Refresher: AWS Global Infrastructure
Benefits
- 👍 Decreased latency
  - Users closer to infrastructure have lower latency
  - If infrastructure deployed in multiple areas, more people will be closer to infra
- 👍 Resilience and Disaster Recovery (DR)
  - If part of the infra goes down, other regions are still up → app can still operate
  - Attack protection → distributed global infrastructure harder to attack
Disadvantages
- 👎 Increased infrastructure costs
  - More infrastructure means paying more
- 👎 More configurations, more management overhead
  - Must use and configure global services, data replication, redundancy, and failovers
💡 In general, it makes sense to make an app global when the performance benefits are needed, or when it's starting to scale considerably in amount of users

Ref: https://www.udemy.com/course/aws-certified-cloud-practitioner-new/learn/lecture/20056112

🔧 Global DNS service
- Refresher: Amazon Route 53 (R53) 101
Can configure routing users to different infrastructure when they query a DNS endpoint
- Great for routing users to closest infra deployment (least latency)
- Great for DR strategy (route to a different resource if main resource is unhealthy)
Some R53 routing policies:
- Simple Routing: no health checks, DNS endpoint always routes to same infra
- Failover Routing: health check on primary resource. If primary resource is unhealthy, route to secondary resource instead → simple DR strategy
- Weighted Routing: direct different amounts of traffic to different resources
  - e.g. 70% of traffic to server A, 20% to server B, and 10% to server C → weights=70, 20, and 10
- Latency Routing: direct users to servers with lowest latency for them