3.3.1. VPCs, Subnets, and Shared VPC
💡 First Principle: Because a VPC is global but subnets are regional, adding a new region to your network architecture means adding a subnet in that region to your existing VPC — not creating a new VPC and connecting it back to the others.
A custom mode VPC requires you to manually create each subnet with its own IP range in the regions you choose (versus an auto mode VPC, which pre-creates one subnet per region automatically — a convenience most production environments outgrow quickly). VPC Network Peering connects two separate VPC networks (potentially in different projects or organizations) so resources in each can communicate via internal IPs, without traffic transiting the public internet. Shared VPC takes a different approach for a single organization: a "host" project owns the VPC and its subnets, while other "service" projects attach to it, letting multiple teams' resources share one network without each team needing to manage networking themselves.
⚠️ Exam Trap: A scenario describing multiple teams, each in their own project, that all need to share a single common network administered centrally is describing Shared VPC specifically — VPC Network Peering connects otherwise-independent VPCs together, which is a different topology than one VPC shared and centrally administered across several projects.
Reflection Question: A company wants five different application teams, each working in their own project, to all use IP address ranges from the same centrally-managed network, without each team provisioning their own VPC. Which feature fits, and how does it differ from simply peering five separate VPCs together?
