The algorithm of block chain consensus is understood from the perspective of traditional service-side development. Why PBFT is two-thirds +1 that 2/3+1,paxos is One-second +1 that 1/2+1__ block chain

Traditional server-side developers may be more familiar with Paxos,zab,raft. Zab and raft are essentially two elections. This is actually a simplified version of the Byzantine problem, without a traitor. But the communication channel could be compromised.
Byzantine error tolerance (BFT B fault tolerance) in a public environment can be disruptive, such as double pay in encrypted currency.
Word, the consensus of the best design is modular, such as notary, the choice of consensus algorithm is highly relevant to the application scenario, the trusted environment using Paxos or raft, with the license of the Federation can use PBFT, the non-license chain can be pow,pos,ripple consensus, etc. According to the Counterparty trust rating, the free choice of consensus mechanism, this is really the best.

PBFT (PBFT is the abbreviation of practical Byzantine Fault tolerance, meaning the practical Byzantine fault-tolerant algorithm): Consensus efficiency at least two aspects, one is to confirm the time, First, the number of scalability--network nodes and even dynamics. PBFT is good at permissioned setting, but his communication is at least squared, and obviously is hard to support large network nodes, even in permissioned setting.         According to colleague Test, the Hyperledger 100 nodes are about to finish the egg ... PBFT is the abbreviation of practical Byzantine Fault tolerance, which means a practical Byzantine fault-tolerant algorithm. This algorithm is proposed by Miguel Castro (Castro) and Barbara Liskov (Liskov) in 1999, which solves the problem of inefficient original Byzantine fault-tolerant algorithm, and reduces the complexity of the algorithm from the point of reference to the polynomial level, It makes the Byzantine fault-tolerant algorithm become feasible in the practical system application. The paper was published at the 1999 International Conference on Design and implementation of operating systems (OSDI99). Yes, this Loskov is the man who presented the famous Richter replacement principle (LSP), the 2008 Turing Prize winner. [2] Why PBFT is two-thirds +1 or 2/3+1,paxos is One-second +1 is 1/2+1.  The reason is that 1/3 of the pbft should be considered untrustworthy, so the 2/3+1 must be returned. Phil himself figured it out. [1]


Author: Li Qire
Link: https://www.jianshu.com/p/fb5edf031afd
Source: Jianshu
Copyright belongs to the author. Commercial reprint please contact the author to obtain authorization, non-commercial reprint please indicate the source.


[0] Ask technology Daniel, PBFT is not the best consensus mechanism in the block chain at present. https://www.zhihu.com/question/55794026
[1] Reply: If a node receives a 2f+1 commit message, it can submit a new block and its transactions to the local block chain and state database. Https://www.jianshu.com/p/2383c7841d41
[2] block chain core technology: Byzantine consensus algorithm pbft HTTPS://WWW.JIANSHU.COM/P/FB5EDF031AFD
[3] PBFT algorithm https://www.jianshu.com/p/2383c7841d41