Last updated
In these systems, each entity is given one identifier that solely identifies it as an entity.
Identifiers are strings of fixed size, to which we give the designation of flat names or non-structured.
Flat names can be efficiently manipulated by machines.
Announcing an ID, and waiting for a response from the entity with its current address.
Does not scale beyond the current local network.
Needs that every process listens to requests.
Example:
ARP - Address Resolution Protocol.
When an entity moves, it leaves a reference to its new location.
Eg: MobileIP.
Each node has an m-bit random identifier.
Each entity has a unique m-bit key.
The entity with the k key stays under the jurisdiction of the node with an id that obeys the rule of id >= k.
Each node p keeps a Finger Table with a maximum of m entries.
Search for the k key, the p node redirects the request to the node with index j that obeys the rule:
q = FingerTable[j] <= k < FingerTable[j+1]
Attribution of nodes based on the topology: When an ID is attributed to a node, ensuring that the nodes nearest to the ID domain are also near the network topology is necessary.
Proximity routing: Keeping more than one successor and routing to the one nearest.
Neighbor selection based on proximity: When needing to choose between neighbors, choose the one closer.
