Does it scale?
With the ongoing global COVID-19 crisis, many people need to start working from home. Typically, the use of a VPN is required to be able to keep working. Many current VPN solutions used by institutions can not cope with this situation, either due to hardware, software or license restrictions. eduVPN is a credible solution which does not suffer from these limitations and scales with your hardware, i.e. CPU and network capacity.
Most organizations start by deploying a single server, which can scale quite well to around 1000 simultaneously connected clients assuming at least 16 CPU cores with AES-NI and adequate network performance, e.g. >= 10 Gbit interface(s).
But it is also possible to deploy extra servers with OpenVPN processes in order to allow for a higher number of concurrent users, or distribute over different locations. Per server (with 16 cores and >= 10 Gbit) another 1000 clients can connect. (However, just as with other VPN solutions this assumes no other restrictions are in place that may limit connection speed, e.g. NAT and/or firewalls.)
When using multiple servers, there is a distinction between controller and node(s). The controller runs the portal and API, the node runs the OpenVPN process(es).
- Machine 1 has both controller and node functionality in location X;
- Machine 2 has node functionality in location Y;
- Machine n has node functionality in location N.
In order to securely add node(s) to your VPN setup we implemented a simple VPN daemon that runs on the node(s). The communication channel between the controller and node is protected by TLS (client certificates) when contacting remote nodes.
You can also deal with scaling up by assigning different profiles to your end-users, allowing for example split-tunneling for certain users in order to offload your infrastructure and to protect specifically the traffic between their machines and the resources at the institution.
Contact us if you have any questions!