VNS3 LNKe: Creating Cloud-Agnostic Transitive Networks Without a lot of Fuss

by | 16 Dec 2021

Cohesive Networks has been helping our customers build robust transit networks on public cloud infrastructure since our early days. Doing so on VNS3 technology gives you secure and observable methods consistent across cloud providers and other virtualization platforms. Up until recently we achieved this by creating site to site IPSec tunnels into our federated mesh backbone. This approach, while robust due to BGP failover capabilities, adds quite a lot of complexity. Each of these connections have unique peering addresses and autonomous system numbers (ASN), as well as peer access lists to configure and manage. Which brings us to our new offering, the VNS3 LNKe controller. LNKe controllers are simple to set up while still providing robust failover capabilities.

The VNS3 LNKe controller is one of Cohesive Networks latest offerings. It’s been designed to provide a low cost, easy to deploy, method of connecting your private cloud networks, regardless of the provider. Let’s take a look at the mechanics of it.

VNS3 can be deployed in a peered mesh topology, where by all of the members of the mesh exchange connection and routing information with all of the other members of the mesh across encrypted peering links. These mesh peers can be situated in any cloud provider and in any region. This is the hub in your typical hub and spoke model. The difference being that VNS3 hub, or mesh, components can exist in many different locations, while still being aware of all of the other components. Extending the hub
simply entails adding new peers. This hub can be as little as one or two VNS3 controllers to many tens of controllers spanning across your cloud vendors regions. Within this mesh you have full visibility and attestability of network flows.

    Now to connect your various networks into the mesh so as to facilitate your transitive network. LNKe is a light weight variant, thats has been designed to work with the encrypted overlay networking capabilities of VNS3. It uses the cryptographic key architecture to create a tunnel from the LNKe controller to the closest mesh controller. This link can be established through a VPC peering link between the connecting VPC or over public IP. You simply have to deploy the LNKe controller into the connecting VPC and push the VNS3 client pack to it. This gives it a unique overlay address that the hub mesh is aware of.

    The LNKe can be configured to have failover hub members that it will connect to should any failure occur. On the hub members that it is configured to connect to we then create route entries for the LNKe’s network. This route is pointed at the overlay IP that has been associated with the LNKe controller. While these are effectively static entries, VNS3 will only ever enable the one that is actively connected to. We call this dynamic static routing.

    On the connected VPC you can set your subnet route of 0.0.0.0/0 to point to the LNKe controller, since LNKe can also serve duty as your NAT gateway. In this way any traffic that is bound for other connected networks will traverse into the hub, where as non transit network traffic can get out as needed.

    This solution gives you a lot of flexibility in managing your network connections. You have full firewall capabilities to restrict and shape traffic. You can transform traffic should you have overlapping CIDRs. You can combine other connections into the mesh such as remote workforces or data center connectivity. You can inject network function virtualization like NIDS and WAF. You end up with a network control plane that works the same across all cloud providers that is cost effective and easy to deploy and mange.