Unique node lists (UNL)

“Collectively trusted sub-networks” consensus algorithms called U NL are

used by Ripple to deal with high latency, which usually characterizes BFT-

tolerant systems. In order to reach a consensus, a node requires to ask its

own U NL in place of the entire network. This mechanism allows less than

one-fifth of its nodes to be faulty.

Stellar

This is an open Byzantine agreement consensus protocol that is resistant to

Sybil attacks. It solves the problem of building consensus in the presence of

Byzantine nodes without relying on any computational inefficient methods

like PoW. It uses the concept of quorums and quorum slices. A quorum is

defined as a set of nodes required to reach an agreement. A quorum slice is

a subset of a quorum that can convince another specific node to agree.

Quorums have to intersect for building global consensus in the entire

system.

Paxos

It is a popular, fault-tolerant, asynchronous, distributed consensus algorithm

comprising of processes that can propose values. To ensure only a single

value from the set of proposed values is chosen and must be learned by

everyone is the obj ective of the algorithm. The nodes in Paxos are of three

types: proposers, acceptors, and learners. Any of these roles at any instant

of time can be taken by any system process in the network. Proposers

propose values that should eventually be chosen by consensus. Acceptors

form the consensus and accept the values. Learners are a valuable source of

information because they learn which value was chosen by each acceptor,

and therefore, the consensus. The acceptors either rej ect a proposal or agree

to it and make promises on what proposals they will accept in the future,

ensuring that only the latest set of proposals will be accepted. In any given

implementation of Paxos, a system process can take multiple roles. The

Paxos protocol phases are prepared and accepted.

Raft