Caerus: NIMBLE Task Scheduling for Serverless Analytics
https://www.usenix.org/conference/nsdi21/presentation/zhang-hong
Lazy: Barrier between map and reduce stage
Cost: aggregated duration of all the tasks
Pipeline breaker: an operator that produces the first output only after all its input has been processed
Map: one step
Reduce:
Read all mapper output
Produce reduce operations
Mappers produce, reducers get at the same time
PreviousOwnership: A Distributed Futures System for Fine-Grained TasksNextShip Computer or Data? Why not both?
Last updated