TABLE OF CONTENTS One of the more surprising features of IPv6 was the inclusion of an entirely new communication model, anycast, to join the existing unicast and multicast models. Where unicast communication permits the transmission of packets to one specific node, and multicast permits transmission of the same packet to one or more nodes, anycast adds the ability to send a packet to any one and only one of a group of one or more destination nodes.
Although first specified in the IPv6 specification in 1995 (RFC 1884, IP Version 6 Addressing Architecture ), the anycast concept had been around before then. However, despite the longevity of the idea, as of 2003 there has been very little experience with anycast and related applications.
Starting with an overview, including a brief look at the history of anycast, this chapter introduces the current state of the IPv6 anycast specification, followed by summary of some of the work that is currently being undertaken to make anycast work and to exploit anycast for useful applications.
An experimental specification for IPv4 anycasting was defined in RFC 1546, Host Anycasting Service, in 1993. That document suggested that another address class could be carved out of the IPv4 address space for anycast addresses. In that way, nodes could identify anycast packets simply by looking at their destination addresses and treating them appropriately.
The motivation behind the RFC 1546 anycast experiment was to provide nodes with a simpler way to reach any of a group of interchangeable application servers. For example,...
