Anycast is a network addressing and routing scheme whereby data is routed to the "nearest" or "best" destination as viewed by the routing topology.

The term is intended to echo the terms unicast, broadcast and multicast.

• In unicast, there is a one-to-one association between network address and network endpoint: each destination address uniquely identifies a single receiver endpoint.
• In broadcast and multicast, there is a one-to-many association between network addresses and network endpoints: each destination address identifies a set of receiver endpoints, to which all information is replicated.
• In anycast, there is also a one-to-many association between network addresses and network endpoints: each destination address identifies a set of receiver endpoints, but only one of them is chosen at any given time to receive information from any given sender.

On the Internet, anycast is usually implemented by using BGP to simultaneously announce the same destination IP address range from many different places on the Internet. This results in packets addressed to destination addresses in this range being routed to the "nearest" point on the net announcing the given destination IP address.

In the past, Anycast was suited to connectionless protocols (generally built on UDP), rather than connection-oriented protocols such as TCP that keep their own state. However, there are many cases where TCP Anycast is now used, including on carrier networks such as Prolexic.

With TCP Anycast, there are cases where the receiver selected for any given source may change from time to time as optimal routes change, silently breaking any conversations that may be in progress at the time. These conditions are typically referred to as a "pop switch". To correct for this issue, there have been proprietary advancements within custom IP stacks which allow for healing of stateful protocols where it is required. However, without any technology to heal pop switches, systems like GeoDNS are more appropriate.

For this reason, anycast is generally used as a way to provide high availability and load balancing for stateless services such as access to replicated data; for example, DNS service is a distributed service over multiple geographically dispersed servers.