Why does the pbft algorithm require N > = 3f + 1?
or what happens when the primary node is a fault node?
Why does the pbft algorithm require N > = 3f + 1?
or what happens when the primary node is a fault node?
due to the order in which messages arrive, it may occur that f problem nodes return messages earlier than f normal nodes, but to ensure that more messages are normally received than those with problems, it is necessary to satisfy N-f-f > f = > N > 3f, so at least 3f+1 nodes.
Pbft algorithm tolerates Byzantine nodes and failure nodes, and Byzantine nodes are nodes that may forge messages.
the Pbft algorithm itself is a majority consistency algorithm, that is, most nodes think it is correct if they agree with each other.
assume that f nodes are failed nodes and f nodes are Byzantine nodes, then the correct node needed to reach an agreement is at least one Fidel1. Of course, this is the worst-case scenario. If the set of failed nodes and the set of Byzantine nodes are duplicated, there is no need for a correct node, but in order to ensure that the worst-case algorithm can work properly, it is necessary to ensure that the number of correct nodes is 1.
and the total number of cluster nodes is 3f+1.