In one company I set up a geo-distributed network with several sites and providers.
The diagram shows a simplified and anonymized version of the network.
ISP1 - Autonomous System of the 1st provider
ISP1 Forward Network - bridge network to the 1st provider
ISP2 - Autonomous System of the 2nd provider
ISP2 Forward Network - bridge network to the 2nd provider
AS NET - Autonomous System of organization, size /24
AS Subnet1 - the 1st subnet from the organization's Autonomous System, size /25
AS Subnet2 - the 2nd subnet from the organization’s Autonomous System, size /25
Internal Forward Network - a bridge network between MikroTiks with addresses from a private range.
High availability access to Internet is ensured using BGP.
Let's consider the case when the ISP2 provider fails:
In this case Server2 will have access to the Internet through ISP1. The traffic will be routed through MikroTik2 and MikroTik1.
The problem is that in this case MikroTik2 won't have access to the Internet. This is due to the fact that the source IP will have private address from the "Internal Forward Network".
To operate the BGP, each router has a loopback interface with an address from the organization’s autonomous system:
Code: Select all
/interface bridge
add comment="BGP loopback1" name=loopback1