A Dynamic Virtual Datacenter Selection Strategy for Integrated Cloud Service Platform Construction With Multiclouds

Bo Huang,Mengting Sun,Changlai Du,Xiaoqun Yuan

doi:10.1109/access.2019.2956169

Bo Huang, Mengting Sun + Show 2 more

Open Access

https://doi.org/10.1109/access.2019.2956169

Copy DOI

Abstract

The pay-as-you-go model and network virtualization of cloud computing allow micro and small content businesses (MSCBs) who construct their integrated cloud service platforms (ICSPs) with virtual datacenters (VDCs) to serve their end users with low service latency and construction cost. However, designing a flexible VDC selection strategy to meet the demands of MSCBs is a challenging task. To address this problem, a dynamic VDC selection strategy is designed for MSCBs to construct their ICSPs flexibly with the VDCs from different clouds. To this end, three dynamic landmark selection metrics are proposed and applied to construct the network coordinates. Then, a dynamic VDC selection algorithm is presented to determine the locations and service resources of VDCs, which are purchased from different clouds by MSCBs to construct their ICSPs. Based on our VDC selection strategy, a simulator is developed based on our designed experimental framework to evaluate our VDC selection strategy. The experimental results show that compared with previous server placement strategies, our strategy can actively and effectively determine VDCs' locations and allocate service resources for each VDC with less computing consumption and is a practical VDC selection strategy for cloud construction in a multicloud environment.

Highlights

Cloud computing has transformed the entire Internet service industry due to its market-oriented resource allocation model [1], [2], which significantly reduces Internet service providers’ cost to construct their cloud-based service platforms for their Internet services, especially for many latency-sensitive Internet services, such as news, video and publishing services
The results show that 1) the network coordinates constructed based on our landmark selection metrics achieve lower computational and network probing costs with assured estimation accuracy, and 2) the virtual datacenters (VDCs) selection strategy significantly reduces the computing consumption compared with existing cluster-based server placement strategies
This paper aims to address the challenges of VDC selection from different geo-distributed clouds, which should be implemented by micro- and small-content service companies when they use these selected VDCs to construct their integrated cloud service platforms (ICSPs)

Summary

INTRODUCTION

Cloud computing has transformed the entire Internet service industry due to its market-oriented resource allocation model [1], [2], which significantly reduces Internet service providers’ cost to construct their cloud-based service platforms for their Internet services, especially for many latency-sensitive Internet services, such as news, video and publishing services. These data center allocation models cannot yield their expected performances for the scenario of VDC selection in ICSP construction In contrast to these works, the latter aims to construct geo-distributed data center-based service platforms for large-scale network services. Regarding the cluster-based algorithms, these algorithms can actively identify suitable ICSPs’ VDCs, their time complexity or the inaccuracy of the generated server placement strategies cannot satisfy the VDC selection of ICSPs’ construction To address these challenges, a practical VDC selection strategy is proposed for ICSP construction in a multicloud environment in this paper, where the server placement framework of NetClust is applied to reduce the measurement cost and locate VDCs pervasively.

RELATED WORKS

PROBLEM FORMULATION

OBTAINING INTERNET INFORMATION

EXPERIMENTAL DESIGN AND PERFORMANCE ANALYSIS

Findings

CONCLUSION