Data Replication in Large Scale Content Delivery Networks: A Genetic Algorithm Approach

Abstract

Content delivery networks (CDN) bring the content close to the users employing the replicated servers. In this context, main issue is allocation of the content replicas to these geographically distributed replica servers. The allocation goal is to minimize the storage and the delivery costs and at the same time to satisfy service-level agreements. In this paper, a mixed integer programming (MIP) formulation for the problem of the allocation has been proposed and then by utilization of the genetic algorithm (GA) its optimum solutions were extracted. Numerical evaluation results show that the proposed algorithm outperforms others in terms of the storage cost, delivery cost and computation complexity.

An Efficient Content Distribution Network Architecture Using Heterogeneous Channels

“…In (7), R w1 represents the total number of broadcast data access requests at the w 1 layer and α e represents the hit rate of the edge convergence layer;…”

“…Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2020.3037164, IEEE Access Equation 5and (7) are the basis of the layer-to-layer relationship in the hierarchical model, and we can deduce the key parameters related to QoS of each layer: 1) Because the effective data acquisition ratio from the lower layer to the higher layer via WAN is 1 − α wi−1 and α wi via the inside of the edge system, and the transmission path length of the data transmitted through the broadcast network is only one hop, therefore, based on derivation of (1), the total access path length is M axDelay H , as expressed by (8):…”

“…They used different routing rules in the multicast phase to effectively reduce the transmission time, and a dynamic cache update algorithm was implemented using a simulated annealing algorithm to meet SLA constraints. Fatin [7] try to minimize the storage and the delivery costs and at the same time to satisfy SLA, they use genetic algorithm (GA) to extract a mixed integer programming (MIP) formulation for the problem's optimum solutions.With SLA as the constraint, the improvement of cache update real-time was not the fundamental optimization goal of its algorithm. In [8], Alabbasi et al studied the overthe top video streaming system based on hierarchical CDN, and proposed using SDTP (stall duration tail probability) as the optimization target.…”