On the Design of Resilient Multicloud MapReduce

Faced with the challenge of dependability, we have recently explored multicloud solutions to increase the resilience and availability of MapReduce. Based on this experience, we present system design guidelines that allow the scaling out of MapReduce computation to multiple clouds to tolerate arbitrary and malicious faults, as well as cloud outages. Crucially, the techniques we introduce have reasonable costs and do not require changes to MapReduce or to the users' code, enabling immediate deployment.