This article describes how to configure routing to allow Cloud Secure Edge (CSE) users to access resources located at a remote site (Site B) via a Connector installed at a central site (Site A), using an existing Site-to-Site VPN tunnel.
Scenario
Best Practice Recommendation
If you have administrative control over the firewall at Site B, the recommended solution is to deploy a Connector directly at Site B. You can then add this connector to the Service Tunnel. This eliminates the need to route traffic over the VPN and provides better performance and redundancy.
If deploying a connector at Site B is not possible, proceed with the routing configuration below.
Configuration Steps
To route traffic successfully from the CSE Connector at Site A to the resources at Site B, two requirements must be met:
Step 1: Configure Connector Routes
Ensure that the IP subnets for Site B are included in the Connector configuration at Site A. This tells the Connector to tunnel traffic destined for Site B through the Service Tunnel.
Step 2: Configure Return Traffic
The most common point of failure is that Site B receives the traffic but drops the response because it does not recognize the source IP (the CSE Access Tier IPs). You must implement one of the following three options to ensure traffic can return correctly.
Option A: Source NAT (SNAT) - Recommended for ease of use
This method "hides" the CSE source IP addresses behind a local IP address at Site A (such as the firewall's LAN/X0 interface). Site B already knows how to route back to Site A's LAN, so no changes are required at Site B.
Option B: Policy-Based VPN Update
If you use a standard Policy-Based VPN (where "Local" and "Remote" networks are defined in the VPN policy itself), you must explicitly allow the CSE IPs across the tunnel.
Option C: Route-Based VPN (VTI) Update
If you use a Route-Based VPN (Tunnel Interface), traffic is controlled by the routing table rather than policy definitions. Site B needs a static route to send traffic destined for CSE back through the tunnel.
Why this is required: Without this route, Site B will likely route the return traffic for CSE IPs out to its default gateway (Internet), causing the connection to fail because of asymmetric routing.
Important Terminology Note