Nutanix Cloud Clusters (NC2) on AWS Architecture and Deployment
What NC2 on AWS is
Nutanix Cloud Clusters (NC2) on AWS runs the Nutanix Cloud Platform - AOS, AHV, Prism Element, and optionally Prism Central - on AWS bare-metal EC2 instances. The cluster lives inside your AWS account and VPC; you pay AWS for EC2 and networking, while Nutanix licenses and manages the software layer through the NC2 console.
Deployment rules
- Only Nutanix-certified bare-metal instance families are supported (for example i3.metal and i3en.metal). General-purpose or burstable instances cannot expose the local NVMe devices and hypervisor features AOS needs.
- Deployment starts in the Nutanix Cloud Console. You choose region, AZ, VPC, subnets, instance type, node count, and an AWS IAM role. The console creates a CloudFormation stack that provisions ENIs, instances, security groups, and IAM resources.
- Cluster nodes should sit in one AZ so storage replication stays within low-latency, high-bandwidth placement. AOS replicates writes across nodes (RF2 or RF3), so node count and failure tolerance are linked.
- Scaling is done through the NC2 console, not by launching EC2 instances manually. Deleting a cluster must also go through the console so CloudFormation and ENIs are cleaned up.
Worked example
A team wants a 3-node cluster in us-east-1. They pick three i3en.metal instances in one AZ, a VPC with a private subnet that has several free IPs per node, and an IAM role with the NC2 policy. The console deploys the stack; Prism Element becomes reachable on 9440. If the subnet is a /28 with only a handful of free addresses, deployment fails even though EC2 quota is available.
Traps
Assuming any EC2 instance type works; forgetting bare-metal service quotas; launching or terminating instances outside the NC2 console; treating NC2 as a SaaS cluster outside your AWS account.