Abstract
This lecture presented several ways of getting around the famous impossibility of consensus. It first presented how hardware, in particular in the case of shared memory, makes it possible to achieve distributed consensus. The partially synchronous hypothesis was then considered, along with its impact on the possibility of consensus. The implementation of consensus introduced the notions of " leader " as well as " obstruction-free " and " lock-free " algorithms.
Here are the main topics covered during this lecture :
- " The consensus power of hardware " ;
- " The consensus power of time " ;
- " The election of a leader " ;
- " The power of randomness to achieve consensus " ;
- " The distributed random number generator ".
