What is a Gossip Protocol
Computer nodes are not far behind men and viruses when it comes to gossiping. The fundamentals of gossip are the same. It is the way of spreading an information to the people of the same world. When you get to know that your manager has put down her papers, you break this information to your colleague on a smoke break. Your colleague likes the information and passes it on to another colleague while you break this information to another colleague of yours. In no time, every person in the company would be aware of the “gossip” about your manager leaving her job. Actually, “in no time” is incorrect; time taken for everyone to know would be of the order of logarithmic of the number of employees in the company. That’s exactly how computer nodes gossip with each other.
Consider a network of computer nodes. Let’s say, node N1 receives a new information. N1 would then randomly select a peer (say, node N2) and share the information. N1 and N2 would then pick a peer each (say, node N3 and N4) randomly and share the information. The process continues in this fashion till the information is passed on to all the connected nodes. Typically, nodes store the time of information exchange also. In the example above, in the first exchange, N2 would store the details of N1, the information, and the time at which N2 got to know about the information.
Where is Gossip Protocol Used?
Gossip protocol works beautifully in a decentralized network of nodes. It is a decentralized way of information exchange. Rules can be built on these nodes to determine the truthfulness of an information. Let’s say if a network obeying gossip protocol holds a rule that when two-thirds of the nodes return the same information, that information will be considered as the truth. In this process, all the nodes are treated equally. It does not matter if a node is more powerful than its peers. The only thing that matters here is the network bandwidth.