Virtual network gateway consists of two instances in an active-standby configuration. If any disruption happens to the active instance, the standby instance will take over (failover) automatically
It is not possible to directly configure the underlaying VMs that are part of the virtual network gateway. The scalability related settings (to maintain consistent throughput for a specific SKU) are completely managed by Microsoft
Azure firewall
Azure Firewall scales out automatically based on CPU usage and throughput
A default deployment maximum throughput is approximately 2.5 - 3 Gbps and starts to scale out when it reaches 60% of that number
It gradually scales when average throughput or CPU consumption is at 60%
Scale out takes five to seven minutes (when performance testing, make sure you test for at least 10 to 15 minutes, and start new connections to take advantage of newly created Firewall nodes)
Azure Firewall consists of several backend nodes in an active-active configuration
Azure Firewall can scale out as much as you need to accommodate changing network traffic flows, so you don’t need to budget for your peak traffic
Autoscaling is built-in, nothing needs to be configured except selecting the right SKU that supports autoscaling (i.e., Standard_v2, WAF_v2) and enabling auto-scale feature when creating Application Gateway and setting instance count
Minimum instance count: i.e. 1 (can also be 0)
Maximum instance count: i.e. 100 (up to 125)
Application gateway auto-scaling is based on the application gateway CPU usage. As default, if CPU over 80% application gateway will scale out.
Autoscaling takes six to seven minutes to scale out and it scale out to maximum 125 instances