Asymmetric Key-Value Split Pattern Assumption over MapReduce Behavioral Model
Ravi PrakashG, M. Vamsee Krishna Kiran, Saikat Mukherjee · International Journal of Computer Applications · 2014
Actual Quantifiability is a concept in MapReduce that is based on two assumptions: (1) every mapper is cautious, i.e., does not exclude any reducer's key-value split pattern choice from consideration, and (2) every mapper respects the reducer's key-value split pattern preferences, i.e., deems one reducer's key-value split pattern choice to be infinitely more likely than another whenever it premises the reducer to prefer the one to the other.In this paper we provide a new approach for actual quantifiability, by assuming that mappers have asymmetric key-value split pattern about the reducer's key-value utilities.We show that, if the uncertainty of each mapper about the reducer's key-value utilities vanishes gradually in some regular manner, then the key-value split pattern choices it can quantifiably make under common conjecture in quantifiability are all actually quantifiable in the original MapReduce with no uncertainty about the reducer's utilities.