In an NC2 environment using Flow Virtual Networking (FVN), the External Routable Prefix (ERP) is used to identify the address space that can reach external networks. If the ERP is correctly configured in both the Transit and User VPCs but outbound communication still fails, the issue typically lies within the virtual routing table. A static route (such as 0.0.0.0/0) must be manually added to the User VPC routing table with the next hop set to the overlay-external-subnet. Without this explicit static route, the virtual router within the User VPC does not know how to direct traffic toward the Transit VPC for outbound delivery, even if the address ranges are theoretically compatible. Verifying this routing path is a standard step in ensuring overlay workloads can reach on-premises or internet resources.

For a new NC2 on AWS deployment where workload VMs have low storage and memory requirements but high CPU frequency (>3.0 GHz) requirements, and the goal is to minimize the number of CPU cores to reduce application licensing costs, the i3.metal instance type is the most suitable.
i3.metal:
High CPU Frequency: i3.metal instances offer high-frequency Intel Xeon processors (up to 3.1 GHz) which meet the high CPU frequency requirement.
Low Storage and Memory: These instances come with a balanced amount of storage and memory, suitable for workloads with low requirements in these areas.
Minimized CPU Cores: i3.metal instances have fewer CPU cores compared to other high-frequency instances like i4i.metal, making them ideal for minimizing application licensing costs.
Other Instance Types:
z1d.metal: While also offering high CPU frequency, these instances typically come with a higher core count and more memory, which may not be optimal for minimizing licensing costs.
i4i.metal: Designed for I/O intensive applications with higher core counts.
m5d.metal: Balanced instance type but with more cores and not as high CPU frequency as required.
Reference:
AWS EC2 Instance Types Documentation
Nutanix Cloud Clusters on AWS Administration Guide
Nutanix Best Practices for Instance Selection

For VDI workloads that require GPU acceleration, the g4dn.metal nodes are the appropriate choice. These instances are equipped with NVIDIA T4 GPUs, which are well-suited for graphics-intensive applications such as VDI workloads that need GPU acceleration. Other node types like AN36P, m6g.metal, or AN36 do not provide the necessary GPU capabilities.
Nutanix Support and Insights - GPU support in AWS

Successful onboarding of an AWS environment into the Nutanix Cloud Clusters ecosystem requires a precise set of administrative permissions to be present on the account used for initial configuration. Specifically, the AWS user must haveCreateRoleFullAccessto allow the automated scripts to generate the necessary service- linked roles. Additionally, the user must possessIAMFullAccessto manage identity policies andAWSCloudFormationFullAccessto execute the deployment templates. This specific combination ensures that the NC2 portal can correctly configure the " Cross-Account Role " which serves as the primary communication bridge between the Nutanix management plane and the AWS infrastructure APIs. Failing to provide any of these three core permissions will result in an " Insufficient Permissions " alert during the cloud account registration process.

When reserving licenses for Nutanix Cloud Clusters (NC2), administrators must understand the inclusion of add-ons within different tiers. For organizations requiring multi-protocol storage, the Nutanix Unified Storage (NUS) add-on is flexible and can be associated with either the NCI Pro or NCI Ultimate licensing tiers.
Furthermore, the NCI Ultimate tier is designed to be comprehensive; it automatically includes high-value features such as Advanced Replication and Security (Microsegmentation) by default. This simplifies the licensing process because these features do not require separate manual applications once the Ultimate tier is selected. While NCI Ultimate is feature-rich, it does not include the NUS add-on by default, which remains a capacity-based selection that must be specifically managed alongside the base license tier.