Container Resource Allocation for Delay-Guaranteed Service Function Chains in Dynamic Environments

Network function virtualization (NFV) enables implementing a network service as a service function chain (SFC) composed of virtual network functions instantiated in virtual machines or containers on servers. Due to the increasing dynamics in resources and traffic in networks, traditional static paradigms for SFC placement and resource allocation are inadequate to meet the delay requirements of emerging services. Motivated by advancements in containerization technologies, particularly vertical scaling of containers, we propose a lightweight online framework for container resource allocation, aiming to maximize the service completion rate while providing delay guarantees in dynamic environments. Our proposal manages resource allocation among multiple containers on servers by bounding physical backlog queues and constructed virtual queues. Numerical results demonstrate that the proposed algorithm achieves a superior service completion rate compared to a static allocation method and a backpressure algorithm.