Anycast routing lets multiple locations advertise the same IP prefix. The internet routes each client toward one reachable instance based on BGP path selection, not a central decision maker.

That makes anycast a strong fit for stable global entry points: public DNS resolvers, CDN edges, DDoS scrubbing, and API front doors. BGP routing is a network-layer decision, though. It has no view of server load, and "nearest" means topologically preferred along the chosen path rather than physically closest.

Before anycast makes sense, it helps to compare it with the other common IP delivery models.

1.1 Unicast: One Sender to One Destination

Unicast is the normal model for most network traffic. One destination IP address identifies one reachable endpoint, or one load balancer VIP, at a particular place in the network.

When a browser connects to an application server, it usually uses unicast.

1.2 Broadcast: One Sender to Everyone on a Local Segment

Broadcast sends traffic to every host on the local network segment. It is useful for local discovery protocols such as ARP and DHCP, but routers do not forward general broadcast traffic across the internet.

1.3 Multicast: One Sender to a Group

Multicast delivers packets to receivers that joined a multicast group. It is used in some controlled networks for media distribution, market data, and infrastructure protocols. It is not the default way public internet applications reach users.

1.4 Anycast: One Address, Multiple Reachable Instances

With anycast, multiple locations advertise the same IP prefix, and the client's network picks one path based on BGP. From the client's point of view, it is sending packets to a single address. From the routing layer's point of view, that address has many possible origins.

Anycast is sometimes described as routing to the "nearest" server. More precisely, it routes to the best BGP path from that part of the internet. That often correlates with low latency, but not always.

The important difference is who makes the first routing decision: