Everyone focuses on the "easy deployment" and zero-trust buzzwords. No one talks about the cost explosion when you scale this across AWS, GCP, and Azure.
Biggest mistake: deploying the same way in each cloud. You'll pay 3x for egress and compute.
* **Don't spin up the Perimeter 81 gateway VMs in every region.** You pay for the instance + the outbound data transfer. Map your actual user density and deploy gateways only in your primary cloud or a central region with cheap egress.
* **Use cloud-native Private Link/VPC Service Endpoints (AWS PrivateLink, GCP Private Service Connect) FIRST.** Route internal traffic to your cloud services directly through the cloud backbone. Only send internet-bound or cross-cloud traffic through the Perimeter 81 tunnel. This cuts egress costs by 70-80%.
* **Autoscale the gateways aggressively.** Their default scaling is conservative. You're paying for idle compute. Set aggressive scale-in policies.
* **Track egress costs by gateway location religiously.** The bill won't itemize it nicely. You need tags or a separate cloud cost dashboard to see which gateway is generating the most cross-cloud traffic costs.
If you don't have a FinOps process to monitor the additional egress and compute, don't deploy it multi-cloud. You're just adding a new, opaque layer of cost.
show me the bill