Active/Passive HA Model
Use VM-Series firewall belongs manage instance groups across separate zones using
load balancers.
| Where Can I Use This? | What Do I Need? |
- Google Cloud Platform (GCP)
|
- VM-Series License (PAYG or BYOL)
- VM-Series plugin
- Panorama
- Panorama plugin for GCP
|
In the active/passive HA model, each VM-Series firewall belongs to an unmanaged
instance group across separate zones within the same region. Only the primary (active)
VM-Series firewall receives network traffic from
Google Cloud load balancers. The health check
configured on the load balancers determines the HA state of the primary VM-Series
firewall. If the health check fails on the primary VM-Series firewall, the load
balancers use
connection tracking to carry the active
sessions to the secondary VM-Series firewall. At that point, the secondary VM-Series
firewall becomes the primary firewall while maintaining session continuity.
This model is suited for environments with any or all of the following
requirements:
Maintaining session continuity through stateful failover between the
VM-Series firewalls.
Horizontal scaling of the VM-Series for performance is not
required.
Inability to centrally manage the firewalls using Panorama or Strata
Cloud Manager.
Establishing IPsec tunnels through the load balancers to the VM-Series
firewall.
Preserving the original client IP address for internet inbound traffic
to internal applications protected by the VM-Series firewalls.
For more Information, see:
After deploying the VM-Series firewall, the load balancer's health
checks will only pass on the active VM-Series firewall. This is because the
dataplane of the passive firewall is inactive and is unable to pass the
health checks. During a failure event, when the passive firewall becomes
active, along with its dataplane, the health checks will pass.
Unlike the autoscale model which uses a Cloud NAT or public IP
address, traffic outbound to the internet from the trust VPC uses a
forwarding rule on the external load balancer as the translated address.