Skip to main content
Each Azure Firewall carries a fixed hourly deployment charge, however much traffic it handles. When one region has several production firewalls, their virtual networks can often be peered to one hub and protected by a single firewall. nOps estimates the saving from removing each extra firewall, after the cost of the VNet peering traffic that replaces it.

Why it matters

Firewalls are often added one per team, subscription or project. The deployment charge is the largest part of their cost, and data processing simply moves to the shared firewall.

How nOps finds it

nOps reads the last 14 complete days of pay-as-you-go Azure Firewall billing. It looks for regions with at least two production firewalls deployed for the full day on at least 13 of the 14 days. Non-production firewalls and firewalls in a secured virtual hub are not counted. The firewall kept is the one with the highest SKU, then the highest cost (costs within 2% count as equal), then the most data processed. For each other firewall, the saving is its deployment cost over the last 7 days, minus the intra-region VNet peering list price (in and out) for the data it processed, scaled to a month. It must be at least $5 a month.

Before you start

A hub-and-spoke design you can move the virtual networks into, and permission to change peering, route tables and firewall policies.

How to implement

  1. Confirm the firewalls can share one hub: no separate tenants, regulatory separation or overlapping address spaces.
  2. Move the rules of the firewall being removed into the kept firewall, ideally through a shared Firewall Policy.
  3. Peer each virtual network behind the removed firewall with the kept firewall’s hub network.
  4. Update the route tables (user-defined routes) so traffic goes to the kept firewall’s private IP, and test.
  5. When the removed firewall shows no traffic, delete it, or run az network firewall delete --resource-group MY_RG --name MY_FIREWALL.
  6. Update Terraform, Bicep or ARM templates to match.

Risks and considerations

  • A shared firewall is a shared failure point and shares one set of rules. Keep separate firewalls where isolation matters.
  • Check that the kept firewall has the throughput and SKU features the moved traffic needs.
  • nOps cannot see your network topology or rules from billing, so treat the recommendation as a starting point for design.
  • To roll back, deploy a firewall again and point the routes back to it.

FAQs

Data processing moves to the kept firewall at the same or a lower rate. The new cost is VNet peering for that traffic, which nOps subtracts from the saving.