3.3.3. Network Connectivity and Load Balancers
💡 First Principle: Google Cloud offers several distinct ways to connect networks together, and each one trades off cost, bandwidth, and whether traffic crosses the public internet at all — the right choice depends entirely on how much bandwidth is needed and how sensitive the connection is to using shared internet infrastructure.
Cloud VPN encrypts traffic over the public internet between an on-premises network (or another cloud) and a Google Cloud VPC — lower cost, but subject to internet variability. VPC Network Peering connects two VPCs' private IP spaces directly. Cloud Interconnect provides a dedicated physical (or partner-mediated) connection bypassing the public internet entirely, for the highest bandwidth and most predictable latency, at correspondingly higher cost. Load balancers distribute traffic across backend instances or services, and choosing among them depends on traffic type (HTTP(S), TCP/UDP), scope (global versus regional), and backend type. Network Service Tiers — Premium (traffic routed over Google's private global backbone) versus Standard (routed more like typical internet traffic, cheaper but less consistent performance) — let you trade a small amount of latency/reliability for a lower price on egress-heavy workloads.
⚠️ Exam Trap: A scenario emphasizing "the highest possible bandwidth, most consistent latency, and no public internet exposure" for connecting an on-premises data center is pointing at Cloud Interconnect specifically, not Cloud VPN — Cloud VPN is the answer when the scenario instead emphasizes lower cost and faster setup over raw bandwidth guarantees.
Reflection Question: A cost-conscious startup needs basic encrypted connectivity between its office network and its Google Cloud VPC, without especially demanding bandwidth or latency requirements. Which connectivity option fits best, and what would justify upgrading to Cloud Interconnect later?
