ParaDSP: Parallelism-aware Dynamic SFC Placement in NFV-based Networks
Venkatarami Reddy Chintapalli, Rajat Partani, Balaprakasa Rao Killi, Bheemarjuna Reddy Tamma · 2025
In the context of Network Function Virtualization (NFV), the shift from hardware-based middleboxes to software-based Virtual Network Functions (VNFs) on commodity servers promises cost savings and operational flexibility. However, this transition introduces challenges, particularly the increased end-to-end (E2E) delay of Service Function Chains (SFCs) due to VNF processing overhead from the virtualization layer. The E2E delay further scales with the length of the SFC, posing a critical issue for latency-sensitive applications. To address this, parallelized SFCs (PSFCs) have been proposed, leveraging the parallel execution of VNFs to reduce E2E delay. In this paper, we introduce a novel PSFC deployment scheme aimed at enhancing network performance by effectively utilizing vertical scalability, which involves using idle resources within a server to adapt to fluctuating traffic. A mathematical model is developed for PSFC deployment that optimizes resource utilization, considering peak traffic rates of PSFCs, and manages accepted SFC requests while accounting for virtualization overhead. We also propose a heuristic algorithm paraDSP, which increases vertical scalability and ensures E2E delay constraints of PSFCs. Simulation results demonstrate that the proposed scheme significantly improves performance in terms PSFC acceptance rate and packet loss compared to state-of-the-art solutions.