Instant Connect Enterprise can be deployed onto a Kubernetes cluster that is either physically located at a customer's site ('on-prem') or onto a Kubernetes cluster running in a public cloud, like Amazon Web Services or Microsoft Azure. Although the installation procedure is different, once installed, Instant Connect operates identically no matter the environment. The choice of which to utilize depends on the needs of the organization.

When deploying in a public cloud, we recommend utilizing a managed Kubernetes service (such as Amazon EKS, Google GKE or Azure AKS) in which Kubernetes is provided as a service and the administrator is freed of the burden of managing the control plane elements of the cluster. While not advisable, it’s possible to construct your own cluster using raw cloud compute resources (like Amazon’s EC2 instances). Most organizations will achieve best results utilizing a managed service where networking, ingress, and load balancing concerns have been addressed by the provider.

Some elements of your Instant Connect system will likely need to remain on-premises even when the ICE Server is hosted in a public cloud: ICE Telephony should be deployed on-prem, adjacent to your organization’s call manager. Similarly, donor radios linked to ICE channels will require a Static Reflector deployed in the same multicast domain as the radio gateway (also on-prem).