Exercises › Multi-Region Deployment Language › Global Capacity Planning Vocabulary

Global Capacity Planning Vocabulary

5 questions · Multi-Region Deployment Language

1. An architect says: "We need N+1 region redundancy for our active-active deployment." With three regions currently handling the load, what does N+1 require?

A fourth standby region that activates only during failures, running at minimal cost Each of the three active regions must have enough capacity to absorb the full load if one region fails — so each region runs at no more than 50% capacity during normal operations with two of three regions able to handle 100% of traffic An additional monitoring region that tracks capacity metrics globally

2. During an incident, an SRE announces: "We're shifting 20% of traffic to eu-west-1 to reduce load on us-east-1." What mechanism makes this possible?

Database replication weights that cause 20% of writes to go to eu-west-1 A global load balancer or DNS routing policy with adjustable weighted routing rules An autoscaling rule that automatically moves 20% of running instances to the other region

3. A platform engineer distinguishes between a global load balancer and a regional load balancer. What is the key difference?

A global load balancer handles more requests per second; a regional one handles fewer A global load balancer routes traffic between regions, considering geography and region health; a regional load balancer distributes traffic between instances within a single region A global load balancer uses Layer 7 HTTP routing; a regional load balancer uses only Layer 4 TCP routing

4. A capacity planner says: "Each region must be able to handle burst traffic independently." What does this requirement mean for provisioning?

Each region needs spare capacity above its normal baseline to absorb sudden traffic spikes without relying on other regions Burst traffic is handled by the CDN layer, so regional capacity planning is not affected The requirement means adding a burst region that only activates during traffic spikes

5. An engineering manager reports: "We've achieved regional isolation in our architecture." What does this mean for fault containment?

Each region uses a separate VPC that cannot communicate with other regions, making cross-region features impossible Failures, deployments, and configuration changes in one region cannot cascade to affect other regions — each region fails and recovers independently Regional isolation means each region has its own engineering team responsible for operations

Exercise complete!

out of 5 questions

← Back to Multi-Region Deployment Language All exercises