DEFT: Distributed, Elastic, and Fault-tolerant State Management of Network Functions

Abstract

Network function virtualization is the key to developing elastically scalable and fault-tolerant network functions (e.g. load balancer, firewall etc.). By integrating NFV and SDN technologies, it is feasible to dynamically reroute traffic to new network function (NF) instances in the event of an NF failure or overload scenario. The fact that the majority of network functions are stateful makes the task more challenging. State migration and state replication are common approaches in achieving elasticity and fault tolerance. The majority of the studies in the literature either emphasize fault tolerance or elastic scalability while designing a state management system for network functions. In this paper, we have designed a complete state management system, called DEFT, that meets both elasticity and fault-tolerance goals. Our system also supports strong consistency on global state updates. While existing designs rely on a central controller or remote central storage to achieve strong consistency on state updates, DEFT utilizes distributed consensus mechanism to achieve the same. We have done a proof of concept implementation of DEFT and extensively tested DEFT under several model conditions to evaluate its scalability and performance. Our experimental results show that DEFT is scalable and maintains a considerably high throughput throughout. It incurs minimal performance overhead while achieving strong consistency on state updates.