Cost And Energy-aware Virtual Cloud Data Center Embedding Algorithms

With the development of cloud computing technology, more and more critical applications which based on cloud computing appear continuously in people’s life. Virtualization provides a good technical support for the implementation of cloud computing. One critical issue in CDC network virtualization technology is virtual data center (VDC) embedding. When node and link constraints (including CPU, memory, network bandwidth, and network delay) are both taken into account, the VDC embedding problem is NP-hard, even in the offline case. Node heterogeneity and CDC network scale bring challenges to the VDC embedding. Existing algorithms either consider the embedding of nodes or links only, or consider the energy efficiency or the effective utilization of resources only.This paper aims to explore a VDC embedding approach with the objective to produce CSP’s high long-term revenue in heavy-workload CDCs, while minimizing energy consumption in light-workload CDCs. We summarize the main contributions as follows:(1) Propose a Nearest-edge-Switch approach to map virtual Servers (NSS). Edge-switches are connected to physical servers directly. NSS reduces energy consumption and embedding cost by using those physical servers, which are as close as possible, to host virtual servers. NSS can not only reduce the complexity caused by the CDC’s large scale and reduce the number of active servers but also implicitly facilitate the embedding cost reduction in virtual switches and virtual links mapping. (2)Propose a Joint mapping approach to map virtual Switches and Links (denoted as JointSL) to minimize energy consumption and embedding cost. Our experiment results indicate that JointSL algorithm can produce perfect results in a reasonable time. To the best of our knowledge, there is no published literature which explored mapping virtual switches and virtual links in an explicitly joint way. (3)Investigate the effectiveness and efficiency of Breadth-First-Search (BFS) algorithm in virtual link mapping. We also investigate two kinds of GBFS algorithms, namely Switch-NSS-GBFS and Server-NSS-GBFS. The only difference between them is that Switch-NSS-GBFS starts from the physical switch when mapping a virtual link, but Server-NSS-GBFS starts from the physical server when mapping a virtual link. (4)Perform a comprehensive comparative study of NSS-JointSL, NSS-GBFS with the existing VDCE algorithms in various scenarios, including different CDC network topologies (single-root, fat-tree, and clos), different sized substrate networks (small-sized, medium-sized, large-sized networks and ultra-large-sized networks), different sized VDCs, and different VDC topologies.