An Optimization Framework for Migrating and Deploying Multiclass Enterprise Applications Into the Cloud

Abstract

Enterprises can reduce the computational burden and costs substantially by migrating and deploying their partial or even whole applications to the cloud, so as to promote and realize their digital transformation. In this article, we study the following problems in the migration and deployment of enterprise applications: i) How the migration time factor influences application migration indirectly? ii) What is the optimal deployment strategy for multiple applications? In this regard, many existing schemes that aim to optimize the economic cost can neither model the optimal migration strategy nor the optimal deployment resource allocation appropriately for enterprise applications. To tackle these limitations, first, this article aims at minimizing migration time by allocating the bandwidth of the access links for applications migration and formulates a strictly convex optimization problem. After that, the article concentrates on modelling the deployment interactions for resource allocation between enterprise application and cloud physical machines as a non-convex optimization problem. The successive approximation method is used to approximate the problem into a series of strictly convex optimization problems and an algorithm is proposed to achieve the optimal resource allocation for applications deployment problem. Numerical results illustrate the effective performance of the proposed schemes of enterprise application migration and deployment in comparison with other methods.