IP Addressing and CIDR

Why this matters

Every VPC, subnet, and peering connection is a CIDR decision you can't easily undo. A /16 gives you 65k addresses but burns a private-IP block; a /24 gives you 254 hosts but caps your autoscaling group before you notice. Multi-cloud peering fails on day one if two clouds picked overlapping ranges.

Demo

Create non-overlapping VPC CIDRs across two clouds and verify the math.

Try it yourself

1. Calculate usable hosts in 10.0.0.0/22 (hint: 1022 in standard, 1019 in AWS) — verify with ipcalc 10.0.0.0/22.
2. Sketch a plan that gives 4 clouds a /14 each from the 10.0.0.0/8 space without overlap.
3. Try creating an AWS subnet at 10.10.1.0/24 inside a VPC of 10.10.0.0/24 — observe the rejection and explain why.

Prompt your AI

Use these three in order. Each builds on the one before.

1. Basics & terminology

In one paragraph, explain CIDR notation like I'm new to it — what does the /N suffix actually mean for the number of addresses?

2. Why it works (the mechanism)

Walk me through how the network mask, network address, and broadcast address are derived from a CIDR like 10.10.0.0/22, step by step.

3. Advanced — application & what's next

Given a multi-cloud topology with 4 clouds and 3 environments each, design a CIDR allocation plan that leaves room for 10x growth and never overlaps.

References

• RFC 4632 — CIDR allocation strategy
• AWS — VPC sizing and reserved IPs