Evaluation of VPNs

Abstract: This thesis evaluated the performance of four different virtual private networks (VPNs): IP security (IPsec), OpenVPN, SSH port forwarding and SSH using virtual interfaces. To evaluate these VPNs, three comparative performance tests were carried out in which the maximum throughput of each VPN was measured. In every test, a specific parameter was varied to observe how it affected the VPNs throughput. The parameters varied were the type of transport layer protocol used, the encryption algorithm used and whether the VPN used compression or not. The results showed, among others, that when TCP traffic was transferred through the VPN and AES-128 was used as encryption algorithm in a Gigabit Ethernet network, the throughput for SSH port forwarding was 168 Mbit/s, 165 Mbit/s for IPsec, 95,0 Mbit/s for SSH using virtual interfaces and 83,3 Mbit/s for OpenVPN. These results are to be compared to the through put measured when no VPN was used, 940 Mbit/s. Three conclusions are drawn from the results of the performance tests. The first conclusion is that the throughput of a VPN depends on the technology the VPN solution is based on, the encryption method that is used and the type ofdata that is sent over the VPN. The second conclusion is that IPsec and SSH port forwarding are the most effective VPNs of the ones compared in this thesis, while OpenVPN and SSH using virtual interfaces are less effective. Lastly, it is concluded that although the different parameters affected the throughput of each VPN, the relation between the VPNs is the same in almost every test. In other words a VPN that performs well in one test performs well in every test.