On the Design and Implementation of Structured P2P VPNs

Centralized Virtual Private Networks (VPNs) when used in distributed systems have performance constraints as all traffic must traverse through a central server. In recent years, there has been a paradigm shift towards the use of P2P in VPNs to alleviate pressure placed upon the central server by allowing participants to communicate directly with each other, relegating the server to handling session management and supporting NAT traversal using relays when necessary. Another, less common, approach uses unstructured P2P systems to remove all centralization from the VPN. These approaches currently lack the depth in security options provided by other VPN solutions, and their scalability constraints have not been well studied. In this paper, we propose and implement a novel VPN architecture, which uses a structured P2P system for peer discovery, session management, NAT traversal, and autonomic relay selection and a central server as a partially-automated public key infrastructure (PKI) via a user-friendly web interface. Our model also provides the first design and implementation of a P2P VPN with full tunneling support, whereby all non-P2P based Internet traffic routes through a trusted third party and does so in a way that is more secure than existing full tunnel techniques. To verify our model, we evaluate our reference implementation by comparing it quantitatively to other VPN technologies focusing on latency, bandwidth, and memory usage. We also discuss some of our experiences with developing, maintaining, and deploying a P2P VPN.