Leaderless BFT SMR
WIP
Making PBFT more robust
This does not only include robustness against leader faults, but also network failure, etc.
Allen Clement, Edmund Wong, Lorenzo Alvisi, Mike Dahlin, Mirco Marchetti
Improve the bad case performance, no optimistic case improvement
https://gyazo.com/276853e453ffbeb54544ebc272e7e288
Yair Amir, Brian Coan, Jonathan Kirsch, John Lane
Pierre-Louis Aublin (Grenoble University), Sonia Ben Mokhtar (CNRS - LIRIS), Vivien Quema´ (Grenoble INP)
https://gyazo.com/1a7828ea46010f2d1709dd701e3081f5
Multi-execution
Gauthier Voron, Vincent Gramoli (University of Sydney)
Chrysoula Stathakopoulou, Tudor David, Marko Vukolic´ (IBM Research)
Allows multiple concurrent leaders to propose batches of requests in parallel, in a sense multiplexing several PBFT instances into a single total order.
Weak leader
Tyler Crain and Vincent Gramoli (University of Sydney), Mikel Larrea (Univ. of the Basque Country), Michel Raynal
Weak coordinator: can terminate even when its coordinator is faulty
Eventual synchrony
No leader
Leslie Lamport
Replace the leaders in an ordinary Byzantine Paxos algorithm by a virtual leader that is implemented using a synchronous, simultaneous Byzantine agreement algorithm Cost of leaderleseness: The time required by a leader agreement algorithm that tolerates$ Ffaulty servers is$ F + 1message delays, which replaces the 1 message delay of a leader simply sending a message.
Marius Poke (University of Stuttgart), Torsten Hoefler (ETH Zurich), Colin W. Glass (University of Stuttgart)
Loïck Bonniot, Christoph Neumann, François Taïani
See also
Others
Blockmania
Constructs a DAG