Difference between revisions of "What is DUNDi"
m (DUNDi moved to What is DUNDi: just because) |
|
(No difference)
| |
Revision as of 11:12, 6 February 2007
Ik ben hier nu even mee bezig, dus informatie is niet echt helder :P
DUNDi (Distributed Universal Number Discovery) is a peer-to-peer system for locating Internet gateways to telephony services. In order to do so, it publishes routes and this routes can be accessed with the use of protocols sucs as H323, IAX(2) or SIP.
From Another DUNDi site:
DUNDi is essentially a trusted, peer-to-peer system for being able to call any phone number from the Internet. DUNDi works by creating a network of nodes called the "DUNDi E.164 Trust Group" which are bound by a common peering agreement known as the General Peering Agreement or GPA. The GPA legally binds the members of the Trust Group to provide good-faith accurate information to the other nodes on the network, and provides standards by which the community can insure the integrity of the information on the nodes themselves. Unlike ENUM or similar systems, DUNDi is explicitly designed to preclude any necessity for a single centralized system which could be a source of fees, regulation, etc.
Much less dramatically, DUNDi can also be used within a private enterprise to share a dialplan efficiently between multiple nodes, without incuring a risk of a single point of failure. In this way, administrators can locally add extensions which become immediately available to the other nodes in the system.
In the DUNDi model, there is no central repository. Instead, node participate in a trust system in which each node has a trust relationship with at least one other node in the system. When the client requests a number or extension it does not know how to terminate, it queries the nodes directly connected to it. Those nodes in turn will query nodes which are directly connected to them and so on (care is taken in the protocol to minimize the actual number of transactions and nodes that are queried, while maintaining the same effective answer as querying the entire system). The resulting answers are collated, cached appropriately at each involved node, and passed along to the original requesting party.
