Software License Consolidation and Resource Optimization in Container-based Virtualized Data Centers

Abstract

Over the past two years, the pandemic-induced demand for resources and services has accelerated. The paradigm of softwarization as well as webization, services, and network virtualization have caused disruptive changes within cloudified environments. Software services are the most crucial part of the evolution of these environments. So far, proprietary solutions are particularly emphasized owing to the current economic climate, such as the numerous cloud services that depend on commercial software. Thus, the ubiquity of this kind of software in the cloud creates a new dimension for software license (SL) optimization. For a prosperous company, optimal use of SL resources remains a capital and essential criterion. Managing these resources, virtualized hardware resources, energy, and costs generated in the cloud is a recent challenge. Although many consolidation strategies have been developed in the area of resource management to optimize standard resources and the resulting energy consumption, this problem remains topical despite optimization efforts. Indeed, appropriate schemes and models for the dynamic consolidation of resources and the optimization of data center states have always been under constant investigation and extensive study and development at the infrastructure level. But, so far, the software service level lacks models and algorithms that offer dynamic optimization of resource allocation and workload consolidation, especially when proprietary solutions are in hand. In this article, we offer suitable heuristics for dynamic software license consolidation (DSLC) and optimization of resources. These tend to reduce costs within a cloud data center and minimize overall power consumption. Our solutions allow us to optimize the monetary costs of the users and offer more flexibility to reduce the infrastructure costs. The experimental studies carried out from small to relatively big data centers, in both homogeneous and heterogeneous resource scenarios, show the efficiency of our solutions. Thus, we recorded a rate of 71.4% energy savings, 79.91% total cost savings, 80.25% license cost, and 76.5% virtual machine (VM) cost. We also highlighted a rate of 78% of deactivated physical machine (PMs) and 76.5% of released VMs.

Appendix: The Numerical Results of our Model

Table 5 Experimental results (best) for different consolidation schemes in homogeneous scenario [total user cost = total VM cost+ total license cost; NO consolidation = the initial DC state]

Table 6 Experimental results (best) for different consolidation schemes for heterogeneous scenario [total user cost = total VM cost+ total license cost; NO consolidation = the initial DC state]