Nutanix NCP-MCI-6.5 Practice Test - Questions Answers, Page 6

A GPU (graphics processing unit) is a specialized hardware device that can accelerate graphics rendering and computation for applications that use APIs such as DirectX, OpenGL, CUDA, and OpenCL. A GPU can also offload the CPU from encoding and decoding tasks for remote display protocols such as Frame Remote Desktop Protocol (FRP). A VM can use a GPU either by directly accessing a physical GPU (pGPU) on the host or by using a virtual GPU (vGPU) that shares a pGPU with other VMs. A user running a computer aided design (CAD) application may benefit from using a GPU or a vGPU to improve the performance and responsiveness of the application, especially when moving windows or rendering images. However, if the GPU or vGPU is not properly configured or provisioned, the user may experience slow response time within the VM. Therefore, to diagnose the problem, the administrator should monitor the GPU Usage metric for the VM.The GPU Usage metric shows the percentage of GPU resources that are consumed by the VM over time3.The administrator can use Prism Central to view the GPU Usage metric for each VM in a chart or a widget4.The administrator can also use Prism Central to view other metrics related to GPU performance, such as GPU Memory Usage, GPU Encoder Usage, and GPU Decoder Usage3. By analyzing these metrics, the administrator can determine if the VM is using the GPU efficiently and optimally, or if it needs more or less GPU resources.

NTPservice needs to be configured to ensure logs, alerts and information is consistent across clusters that are located in different countries.NTP stands for Network Time Protocol and it is used to synchronize the clocks of all the nodes in a cluster1.This helps to maintain accurate timestamps for logs, alerts and other information that are generated by Nutanix clusters1.

Enable Memory Overcommit on the over-provisioned VMsis the most efficient resolution to enable additional VMs to be created.Memory overcommit allows VMs to use more memory than physically available on a host by compressing and swapping memory pages to storage1.This can improve memory utilization and increase VM density on a cluster1.However, memory overcommit is not supported when HA is configured to use reserved hosts, so you may need to disable HA reservation on the cluster before enabling memory overcommit1.

https://hyperhci.com/2019/07/22/nutanix-lcm-upgrade-process-failed-trouble-shooting/

An agent VM is a special type of VM that is used for running services or applications that are not affected by high availability (HA) events or host maintenance. An agent VM is never migrated to any other host in the cluster. If an HA event occurs or the host is put in maintenance mode, agent VMs are powered off and are powered on on the same host once that host returns to a normal state.This behavior is different from regular VMs, which are migrated to another host in the cluster if possible, or restarted on the same host if not possible1.

Therefore, during an AHV upgrade, an administrator may find that a critical VM was powered off rather than migrated to another host if the VM was marked as an agent VM.This can be done either from the Prism web console by selecting the Use this VM as an agent VM option in the Update VM dialog box, or from the aCLI by using the vm.create or vm.update commands with the agent_vm=true parameter2. To avoid this situation, the administrator should ensure that any critical VMs are not marked as agent VMs before starting the AHV upgrade.

Nutanix provides two types of granular RBAC for AHV hosts: category based and cluster based3. Category based RBAC allows administrators to assign roles to users or groups based on categories, which are key-value pairs that can be applied to various entities in Prism Central, such as clusters, hosts, VMs, images, and networks. Categories can be used to group entities by different criteria, such as department, project, environment, or location. For example, an administrator can create a category key named Department and assign different values to it, such as Finance, Marketing, or Engineering. Then, the administrator can apply this category to different clusters or hosts and assign roles to users or groups based on this category.This way, users or groups can have different levels of access to different clusters or hosts depending on their department4.

Cluster based RBAC allows administrators to assign roles to users or groups based on specific clusters registered in Prism Central. For example, an administrator can create a role named Cluster Admin and assign it to a user or group for a particular cluster.This way, the user or group can have full access to that cluster and its hosts and VMs, but not to other clusters5.

https://portal.nutanix.com/page/documents/details?targetId=Nutanix-Security-Guide-v6_7:sec-cluster-rbac-pc-c.html

Active-Active is a bond mode that allows all uplinks in the bond to be used simultaneously for traffic transmission and reception. This bond mode provides load balancing and increased bandwidth for the AHV host and its VMs. Active-Active bond mode uses a hashing algorithm based on source MAC addresses to distribute traffic across different uplinks in the bond. Each individual VM NIC uses only a single bond member interface at a time, but multiple VM NICs are spread across different bond member interfaces.As a result, it is possible for a Nutanix AHV node with two 10 Gb interfaces to use up to 20 Gbps of network throughput, while individual VMs have a maximum throughput of 10 Gbps6.

Therefore, if an administrator needs to optimize for maximum single VM throughput, they should select Active-Active bond mode for their AHV cluster.This bond mode can be configured using Prism Element UI or manage-ovs commands on each AHV host7.No additional configuration is required on the upstream switch side, as long as the switches are interconnected physically or virtually and both uplinks trunk the same VLANs8.

A Nutanix cluster requires a minimum of three nodes to maintain quorum and data availability. Removing a node from a 3-node cluster would violate the redundancy factor and cause data loss. Therefore, it is not possible to remove a host from a 3-node cluster using Prism or CLI.The only way to replace a node in a 3-node cluster is to use the Foundation tool, which will erase the existing cluster configuration and create a new cluster with the new node1.

CloudInit is a package that contains utilities for early initialization of cloud instances. It allows you to customize virtual machines provided by a cloud vendor by modifying the generic OS configuration on boot.You can use CloudInit to set the OS hostname, add custom users, add keys, run custom scripts, and more2.CloudInit is supported by most major Linux and FreeBSD operating systems and works across different cloud platforms3.Sysprep is a tool for Windows operating systems that prepares an installation for cloning, auditing, and customer delivery4.