Networking Beyond Traditional Address Types
So far we’ve discussed EIDs and their mappings in the context of network addresses, but we can define EIDs more broadly to encompass much more than what a network address traditionally conveys.
In the mail system analogy, a person’s identity may be qualified beyond a name to include details such as whether this person should be contacted for work purposes versus personal purposes, or how this person may be contacted after hours or during working hours. There could be further refinement of the identity specifying how to contact the person if the requestor is within the country or if someone is trying to reach the person from abroad. So, the directory may be able to provide the address for an individual based on a more detailed specification of the identity. The notion of identity is therefore extensible, allowing the directory to provide location information in the context of the intended communication policy. Furthermore, the directory may provide information in addition to the street address of the receiver; the obvious example is the phone number of the receiver in addition to street address. Thus, both identity and location naturally have extensible semantics.
In the networking context, the definition of EIDs and RLOCs quickly expands from traditional network addresses to more general data structures capable of incorporating rich information to define communication policy. For example, the EID may be expanded to include information such as the role of the EID (work or personal), a grouping of EIDs, or the time of day. An RLOC, in turn, may be expanded to include grouping information or even geo-coordinates.
The way these extensible types are encoded in the LISP control plane is defined by the LISP Canonical Address Format (LCAF) where a multitude of types is defined to allow the system to operate beyond the traditional network address types of IPv4, IPv6, and Ethernet to include much more flexible semantics capable of encoding composite names or even accommodate for previously undefined namespace types.
The potential of this extensibility is exploited immediately in the context of software-defined networking (SDN)–based applications. For instance, some applications leverage the encoding of geo-coordinates in the RLOC space to leverage the information in the network to enable location-tracking applications for entities that roam around a LISP-enabled network. In this example, the application queries the database for a particular EID, and the Mapping Database System replies to these queries with the IP addresses of the current RLOCs in the mapping. The reply also includes the geo-coordinates of the RLOCs, providing the application with coordinate information it would traditionally have procured from other sources but not from the network.
In looking at what the future may hold, notable research activity is a good indicator of where things may lead. There are efforts to formalize the notion of handling information in the network and make it a core element in the foundation of the Internet. One example is the Information-Centric Networking research group (icnrg) at the Internet Research Task Force (IRTF, the sister research branch of the IETF). This research group is looking at the implications of moving the focus of the Internet from nodes to information objects, which is at a high level moving from networking on IP addresses to names. The group centers its analysis on the use of name-based routing. The motivation and implications behind the work in this group are in line with the motivations and implications behind providing a network directory service capable of supporting extensible address types for both identity and location.