Active/backup configurations and load balancing with Keepalived
A Linux server that is set up as a load balancer for multiple worker servers and keeps the service available, even if any of those workers fail. However, the load balancer itself becomes a single point of failure in that scheme, unless the administrator also takes care to provide a failover mechanism for multiple balancers.
The usual way to achieve failover is by using a floating virtual IP address. Suppose
www.example.com is configured to point at
192.0.2.100. If you assign that address directly to a load-balancing server in a
192.0.2.0/24 network, it becomes a single point of failure. However, if you set up two servers with primary addresses from that network (say,
192.0.2.20), you can use a special failover protocol to allow two or more servers to decide which one will hold the virtual
192.0.2.100 address and automatically transfer it to a different server if the primary server fails.
The most popular...