| 摘要: | 網絡軟件化是基於SDN (Software Defined Networking)和NFV (Network Function Vir-tualization)形式的雲端計算原理的一種範式轉變，目的是即時地提供依照需求、低成本和服務導向的網絡。雖然SDN通過其邏輯集中式架構帶來更多的轉發靈活性，但NFV為其虛擬化技術處理帶來更多靈活性。本文提出了三種不同的解決使用SDN和NFV的網絡軟件成本問題的方法。我們的第一種方法是基於SDN，其他兩種基於NFV。
儘管在SDN網絡中傳輸調度較為單一，但由於分佈式傳統網絡的安裝基數較大，因此SDN節點可能需要與傳統節點共存以形成混合網絡。儘管混合網絡比純粹的傳統網絡具有更好的性能，但由於預算有限，仍然存在SDN快速升級的問題。為了評估混合網絡中成本和覆蓋範圍之間的關係，我們將路徑中SDN節點和至少有一個SDN節點的路徑的最小百分比分別定義為跳數覆蓋和路徑覆蓋。為了評估成本和覆蓋範圍之間的關係，我們將SDN節點選擇分別作為四個跳轉/路徑覆蓋和成本作為目標和約束的優化問題，反之亦然。我們為路徑和跳都制定了SDN覆蓋問題，並提出了兩種算法：1）最大未覆蓋路徑數（MUcPF）和2）最小覆蓋路徑優先數最大值（MMHcPF）。 MATLAB實驗結果表明，與其他現有算法相比，我們提出的算法在覆蓋範圍，成本和效率方面具有更高的性能。
NFV通過虛擬化網絡功能（VNF）鏈將虛擬化擴展到網絡，以便在運行時按需提供所需的功能。這些VNF對其所放置的物理機（PM）的功耗有直接影響。 PM和這些機器的不同負載及其利用率是網絡能耗的關鍵問題。為了解決這個能源問題，我們設計了一個採用NFV技術的動態節能模型，並製定了能源成本優化問題。我們提出VNF鏈（DPVC）算法的動態放置作為解決方案。結果表明，與其他算法相比，我們的算法性能更好，節省更多能源。
電信運營商將其中央辦公室（CO）和移動基站重新設計為NFV數據中心（DC），稱為邊緣數據中心（EDC），可幫助網絡運營商加速部署並降低成本。早些時候NFV的使用被限制在DC內。最近有許多研究正在跨DC使用NFV，即DC間。然而，這些NFV的DC間架構限制了DC之間的通信，無論是水平連接還是垂直連接。我們提出了這些邊緣NFV數據中心的通用體系結構，包括水平和垂直連接，並製定成本優化問題以證明DC之間在通信和計算方面的垂直和水平連接的後果。我們通過估計DC之間的流量調度速率來提出一個帶有延遲和容量約束的成本優化問題，並提出一種垂直 - 水平通信（VHC）啟發式解決方案。所獲得的結果表明，垂直連接與水平連接相比有助於顯著降低計算成本。但是，垂直和水平通信一起可以幫助降低通信成本和總成本，而不是僅垂直或水平通信。
;Network softwarization is a paradigm shift based on cloud computing principles in the forms of SDN (Software Defined Networking), and NFV (Network Function Virtualization), in order to provide on-demand, cost-efficient and service-oriented networks on-the-fly. While SDN brings more flexibility in forwarding by its logically centralized architecture, NFV brings more flexibility in processing with its virtualization technology. This thesis presents three different approaches to address the cost issue of network softwarization using SDN and NFV. Our first approach is based on SDN and other two are based on NFV.
Despite having smatter transmission scheduling in SDN network, SDN nodes may require to coexist with legacy nodes to form hybrid networks due to the large installed base of distributed legacy networks. Although, a hybrid network has better performance than a pure legacy network still rapid up-gradation to SDN is an issue due to the limited budget. To evaluate the relationship between cost and coverage in the hybrid network, we define the minimum percentage of SDN nodes in a path, and paths with at least one SDN node, as the hop coverage and path coverage, respectively. To evaluate the relationship between cost and coverage, we formulate SDN node selection as four optimization problems with hop/path coverage and cost as objectives and constraints, respectively, and vice-versa. We formulate SDN coverage problem both for the path and hop and propose two algorithms: 1) maximum number of uncovered path first (MUcPF) and 2) maximum number of minimum hop covered path first (MMHcPF). MATLAB experiment results demonstrate that our proposed algorithms have greater performance in terms of coverage, cost, and efficiency compared to other existing algorithms.
NFV extends the virtualization to networking by virtualized network function (VNF) chaining to provide the required functionality at runtime on demand. These VNFs have a direct impact on power consumption of the physical machines (PM) on which they have placed. The PMs and varying load to these machines and their utilization are the critical issues of the energy consumption in the network. To address this energy issue, we designed a dynamic energy saving model with NFV technology and formulate an energy-cost optimization problem. We propose a Dynamic placement of VNF chains (DPVC) algorithm as a solution. The results show our algorithm performs better and saves more energy compared to other algorithms.
Telecommunication carriers are re-architecting their central offices (CO) and mobile base stations as NFV data centers (DC), known as edge data center (EDC) that help network operators to speed deployment and reduce cost. Earlier the use of NFV was limited to within the DC. Recently there are numerous studies going on the use of NFV across DCs i.e., inter-DC. However, these NFV inter-DC architectures have limited the communication among DCs to either horizontal connectivity or vertical connectivity. We propose a generic architecture of these edge NFV data centers, with both horizontal and vertical connectivity and formulate a cost optimization problem to demonstrate the consequence of both vertical and horizontal connectivity between DCs in terms of communication and computing. We formulate a cost optimization problem with latency and capacity as constraints by estimating the traffic dispatch rate among DCs and propose a vertical-horizontal communication (VHC) heuristic solution. The obtained results show that the vertical connectivity helps to reduce the computing cost significantly compare to horizontal connectivity. However, both the vertical and horizontal communication together can help to reduce the communication cost and total cost compared to only vertical or horizontal communication. |
