Hierarchical Coded Elastic Computing

Abstract

Elasticity is offered by cloud service providers to exploit under-utilized computing resources. The low-cost elastic nodes can leave and join any time during the computation cycle. The possibility of elastic events occurring together with the problem of slow nodes, referred to as stragglers, increases the uncertainty of the system, leading to computation delay. Recent results have shown that coded computing can be used to reduce the negative effect of elasticity and stragglers. In this paper, we propose two hierarchical coded elastic computing schemes that can further speed up the system by exploiting stragglers and effectively allocating tasks among available nodes. In our simulations, our scheme realizes 45% improvement in average finishing time compared to the state-of-the-art coded elastic computing scheme.

A New Design Framework for Heterogeneous Uncoded Storage Elastic Computing

“…The authors proposed new algorithms using shifted cyclic task allocation to reduce the transition waste and showed it is optimal under some parameter settings. In [3], the authors proposed two hierarchical schemes that can further speed up the USEC system by effectively allocating tasks among available nodes while the encoding and decoding complexity may be increased. Some important limitations of [1]- [3] include the assumption that all available VMs have the same computing speed or the proposed schemes do not consider the heterogeneous computing speed among machines, and all VMs have the homogeneous storage constraint.…”

“…In [3], the authors proposed two hierarchical schemes that can further speed up the USEC system by effectively allocating tasks among available nodes while the encoding and decoding complexity may be increased. Some important limitations of [1]- [3] include the assumption that all available VMs have the same computing speed or the proposed schemes do not consider the heterogeneous computing speed among machines, and all VMs have the homogeneous storage constraint. In practice, based on our own measurement [4], the computing speed among VMs can be significantly different even if they have exactly the same configurations, e.g., same CUP, RAM and I/O cost.…”