Building Connectivity Graph #4
Description
As there is no central point for tracking, each peer will try to find alive hosts in the network. But they cannot keep all found hosts as their neighbors. Hence, it is important to build a graph that can be used for large scale querying, where queries can propagate on the graph. To do that, the current approach planned out is that each peer will try to find alive disconnected hosts. First issue is to determine whether it is from a disconnected because no peer has complete information about the network. Hence, it can query its neighbors which in turn can check if the host is disconnected from the current graph. We do not want it to be perfect. Rather, if the number of hops required to determine that it is indeed disconnected or not is high, then there should be a high chance that it will be assumed to be disconnected.
Once this process is complete, the peer that found the disconnected host can make it its own neighbor or can pass the request to its neighbors. This method is to be used to optimize the graph building for query based work.