An Asset-Controlled Mapping Database
The LISP mapping database naturally embodies the declarative model of object interrelations that is at the heart of modern large-scale software architectures. In a declarative object model, agents are trusted to work autonomously, and they declare what they are willing and able to do instead of receiving a directive from the top-down stating what is expected from them. This capability leads to a scalable distribution of tasks and responsibilities in modern software architectures in which intelligence is a distributed attribute and trust relationships between agents enable the creation of efficient systems.
The concept of intent and declaration is intuitive and not exclusive to software architectures. Much of the social network (no, not the website where you post photos and comments) around which our lives are structured is based on declaration and trust. Simple interactions, such as having a neighbor water your plants while you are out of town, follow a model of declaration of intent. The neighbor declares an intent to water the plants every other day while you are gone and also declares a preference to do this in the mornings and skip the weekend. You, in turn trust, your neighbor to do as promised. There is an implicit assumption in this trust that the neighbor will do his best to take care of the plants and improve his own process of watering and care without your having to know or worry about how he goes on about things. Trusting the neighbor to be competent is important to the success of the system in making society scalable and productive. You, in turn, may be establishing other trust relationships and declaring intent to fulfill other tasks while you travel predicated on the time made available to you thanks to your neighbor’s intent to attend to your domestic business.
In LISP, the ETRs are autonomous and trusted. The ETRs declare their intent to deliver traffic to certain EIDs by registering the EID mappings with the Mapping Database System.
In declaring their intent to forward traffic to the EIDs for which they are authoritative, the ETRs also declare their preferences in terms of how they’d like the traffic to reach them. LISP locators have associated priority and weight parameters that are set by the ETRs. The ETRs are trusted to the extent that they actually control the mapping database through the process of declaration of intent and preference. The ETRs are therefore assets that are listed in the database, and they actively participate in the database and control it by being the authoritative source of information for the database.
The distribution of tasks and state that results from this declarative model is instrumental in making the LISP architecture scalable, efficient, self-documenting, and, to an extent, self-healing. There is also a rather profound operational implication in that administrators of different ETRs can use declarative models to interact with each other and to model their communication policies. Declarative models are critical to the federation and agile definition of communication policies across administrative domains. The declarative nature of the Mapping Database System provides the necessary data model to enable the clean development of programmatic RESTful interfaces for communication with the LISP Mapping Database System and for communication between LISP administrative domains.