Determining Delay Bounds for a Chain of Virtual Network Functions Using Network Calculus

With the development of network function virtualization technology, service flows are expected to be served by a chain of Virtual Network Functions (VNFs). Some works adopt queuing theory to derive the end-to-end delay for a chain of VNFs. However, the delay based on queuing theory is an average delay, and the packets cannot be guaranteed to be transmitted within this delay in the worst cases. In this letter, we apply network calculus to derive the delay bounds for a chain of VNFs. The service flows are prioritized with per-VNF priority, and share the VNF instances. We design and implement a prototype system to validate our derivations, and the results show that the delay bounds derived from network calculus are accurate and tight.