No free lunch! Distributed setting introduces new challenges. Locality (Nodes do not know the whole picture.) E.g.:Distributed MST/SSSP problem. Symmetry-breaking(Nodes can only behave identically.) E.g.:Impossibility of deterministic leader election in anon rings. Distributed systems are NOT fault-tolerant by default. In many cases,clever distributed algorithms are needed. E.g.:Paxos/Raft for the consensus problem. Sometimes,certain level of fault-tolerance is impossible! E.g.:Impossibility of crash-tolerant consensus in async setting. A quick look of fault-tolerance via the consensus problem. No free lunch! • Distributed setting introduces new challenges. • Locality (Nodes do not know the whole picture.) E.g.: Distributed MST/SSSP problem. • Symmetry-breaking (Nodes can only behave identically.) E.g.: Impossibility of deterministic leader election in anon rings. • Distributed systems are NOT fault-tolerant by default. • In many cases, clever distributed algorithms are needed. E.g.: Paxos/Raft for the consensus problem. • Sometimes, certain level of fault-tolerance is impossible! E.g.: Impossibility of crash-tolerant consensus in async setting. A quick look of fault-tolerance via the consensus problem