G)Fault Tolerance Proactive way (Redundant Execution) The problem of stragglers(slow workers) Other jobs consuming resources on machine Bad disks with soft errors transfer data very S⊥OW1y Weird things: processor caches disabled(! When computation almost done, reschedule in-progress tasks Whenever either the primary or the backup executions fir nishes, na t as com leted 25Fault Tolerance 25 • Proactive way (Redundant Execution) – The problem of “stragglers” (slow workers) • Other jobs consuming resources on machine • Bad disks with soft errors transfer data very slowly • Weird things: processor caches disabled (!!) – When computation almost done, reschedule in-progress tasks – Whenever either the primary or the backup executions finishes, mark it as completed