HASP: High Availability SFC Placement Approach in Data Center Network
Ziqi Xu, Qingmian Han, Cheng Bo, Meng Niu, Junliang Chen · 2021
Network Function Virtualization (NFV) is a promising technology that decouples network functions from the hardware. Virtual Network Functions (VNFs) can form service function chains (SFCs) and be orchestrated like internet applications. However, there are higher availability requirements for SFCs because the network's failure causes more severe consequences. This paper proposes the High Availability SFC Placement Approach (HASP) for SFC placement in Data Center. HASP includes a cross-backup (CB) model to improve the availability of SFCs. The software runs on the physical machine (PM), and once the PM fails, the software assigned to it will also fail. The CB model adopts offsite redundant VNF instances to avoid PM failure and guarantee SFC availability. HASP also considers the impact of data link reliability on SFC availability. In addition, a heuristic limited area search optimization (LASO) algorithm is incorporated to place the VNFs of SFCs. The evaluation results show that the LASO algorithm is 300% faster than the traditional heuristic algorithm. Combining the CB model and the LASO algorithm, HASP can improve the availability of SFCs and obtain the optimal placement scheme.