Coded Elastic Computing on Machines With Heterogeneous Storage and Computation Speed

We study the optimal design of heterogeneous Coded Elastic Computing (CEC) where machines have varying computation speeds and storage. CEC introduced by Yang et al. in 2018 is a framework that mitigates impact elastic events, can join leave at arbitrary times. In CEC, data distributed among using Maximum Distance Separable (MDS) code such subsets perform desired computations. However, state-of-the-art designs only operate on homogeneous networks same This may not be practical. this work, based an MDS storage assignment, we develop novel assignment approach for to minimize overall time. first consider scenario computing but then both heterogeneities are present. propose combinatorial optimization formulation solve it exactly decomposing into convex problem find load filling exact assignment. A low-complexity algorithm adapted completed within number iterations equal most available machines.