NA-CLTM cheat sheet

Nutanix Cloud Platform comprises AOS (distributed storage and cluster services), AHV (native hypervisor), Prism (management plane), and optional services such as Flow, Files, Objects, and Era.

Prism Element is the per-cluster management UI and API; Prism Central is the multi-cluster manager that provides categories, RBAC, and cross-cluster operations.

AOS runs as a Controller VM (CVM) on every node; the CVM owns local disks and serves I/O to all hypervisors in the cluster over the internal network.

Cluster services include a distributed metadata ring, Paxos-based configuration consistency, and a curator process that scans for and repairs data placement.

Prism Central requires its own VM deployment and sizing; it is not required for basic single-cluster operation but is required for many features such as categories-based policies and Nutanix Calm.

Nutanix Cloud Clusters (NC2) extends the same AOS/AHV/Prism stack onto public cloud bare-metal instances, keeping the operational model identical to on-premises.

Licensing is per-node and tiered (for example Starter, Pro, Ultimate); features such as Flow networking, Files, or Leap may require specific license levels or add-on entitlements.

Health checks and NCC (Nutanix Cluster Check) validate cluster configuration and are typically run before upgrades or after configuration changes.

Nutanix uses a log-structured distributed file system: writes go to an oplog (persistent write buffer) on the local CVM and are later drained to extent stores as immutable extents.

Data is protected by RF (replication factor) or erasure coding; RF2 tolerates one node or disk failure, RF3 tolerates two, and erasure coding reduces capacity overhead at the cost of CPU and rebuild characteristics.

The metadata ring is distributed across CVMs; each CVM is responsible for a subset of vDisk metadata, and requests are forwarded to the owning CVM via the Stargate/Curator path.

Curator performs background scans that re-protect data after failures, rebalance after node additions, and discard stale extents; ILM (Intelligent Lifecycle Management) decides placement tiers.

The unified cache (formerly Shadow Clones/Content Cache) uses local SSD to cache reads for VMs regardless of which host runs them, reducing cross-network reads.

A cluster can be scaled by adding nodes; data is rebalanced automatically, but a cluster generally should not be expanded beyond the validated node count for the AOS release.

Storage containers define policies such as RF, compression, deduplication, and erasure coding; a vDisk inherits its container's policy at creation time.

AHV datastores map to storage containers; NFS and iSCSI are presented to external clients through the CVM, and volumes can be exposed via Nutanix Volumes.

AHV is a KVM-based Type-1 hypervisor managed through Prism; it uses the Acropolis service for VM lifecycle and the CVM for storage I/O.

A VM's disk is a vDisk managed by AOS; VM snapshots use the same redirect-on-write mechanism as storage snapshots, so snapshot creation is fast and space-efficient.

Acropolis supports live migration of VMs between hosts (with shared storage via the distributed file system) and can perform host maintenance mode by evacuating VMs.

Categories are key-value pairs applied to VMs, hosts, or images; they drive policy-based placement, RBAC scoping, and protection domain membership.

Affinity policies constrain VM placement (for example, keep a set of VMs on the same host or separate hosts) and are enforced by Acropolis during placement and HA events.

Nutanix Guest Tools (NGT) provides in-guest agents for VSS-consistent snapshots, self-service restore, and file-level restore on supported guest OSes.

Images are stored in the image service and can be served directly to VMs without a full copy, enabling fast cloning and reduced storage consumption.

AHV supports GPU passthrough and vGPU configurations depending on host hardware and the AOS/AHV release; verify the compatibility matrix before designing GPU workloads.

Prism Central provides dashboards, alerts, and analysis; Prism Element provides per-cluster views; both expose REST APIs for automation.

LCM (Lifecycle Manager) inventories firmware and software, performs upgrades of AOS, AHV, Prism Central, and firmware, and validates pre-upgrade checks before proceeding.

NCC health checks produce a score and list of failed checks; common failures relate to disk health, CVM resources, time sync, and network configuration.

Alerts have severity levels and can be forwarded via SNMP, email, or syslog; alert policies can suppress or escalate based on categories.

Capacity planning uses effective capacity after RF or erasure coding overhead; Prism reports usable capacity and runway based on historical growth.

Upgrades should be performed with NCC passing, adequate free capacity, and a validated compatibility matrix; AOS and AHV upgrades can be staged and rolled back within supported windows.

Prism Central scaling: multiple Prism Central instances can be registered to manage large environments, and Prism Central should be sized per the official sizing guide based on VM count.

Nutanix Insights (where licensed) provides predictive health and upgrade recommendations based on telemetry; it is separate from on-cluster NCC checks.

Prism exposes REST APIs (v2 and v3) for VMs, clusters, storage, and networking; v3 APIs are the current generation and use JSON payloads with task-based asynchronous responses.

Long-running operations return a task UUID; clients poll the task endpoint until the task reaches SUCCEEDED or FAILED.

Nutanix Calm provides blueprints (application models) with services, dependencies, and actions; blueprints can be deployed across AHV, ESXi, and public clouds.

Nutanix Foundation is used for imaging and cluster creation; it can be driven by a configuration file for repeatable bare-metal provisioning.

The Nutanix Terraform provider and Ansible modules allow infrastructure as code for VMs, images, categories, and networking; providers authenticate via Prism credentials or API keys.

Nutanix APIs support pagination, filtering, and sorting; use the length and offset parameters for large result sets and OData-style filters where supported.

Nutanix Flow provides distributed firewall and microsegmentation; policies are defined centrally and enforced at the vNIC level on each host.

Webhook and event-driven automation can be built by polling task endpoints or by integrating Prism alerts with external orchestration tools.

Prism Central RBAC uses roles, users, and role assignments; roles define allowed actions and can be scoped to categories or clusters.

Local users, LDAP/Active Directory, and SAML identity providers are supported authentication sources; directory integration maps groups to roles.

Categories enable least-privilege scoping: a role assignment scoped to a category value limits the user to objects carrying that category.

Nutanix Flow enforces microsegmentation policies at the vNIC; policies can be applied to categories rather than individual VMs, simplifying change management.

Data-at-rest encryption is supported via self-encrypting drives (SEDs) and software-based encryption; key management can be local or integrated with external KMS.

Audit logs record user and API actions in Prism; syslog forwarding supports centralized log retention for compliance requirements.

STIG and CIS hardening guidance is published for AOS/AHV; compliance scanning can be performed against these benchmarks.

Cluster lockdown mode restricts SSH access to the CVM and requires a supported workflow to disable; it is part of a hardened deployment.

Protection domains group VMs by category or explicit list and define snapshot schedules with retention; they are managed in Prism Central.

Local snapshots are space-efficient and fast but do not protect against site loss; remote replication to another cluster or to a cloud target provides site-level protection.

Nutanix Leap orchestrates DR: it manages protection policies, recovery plans, and one-click failover/failback with runbooks for multi-tier applications.

Async replication sends snapshots on a schedule; RPO equals the replication interval, while RTO depends on recovery plan complexity and target readiness.

Nutanix Mine (or integrated backup vendors) provides backup to external targets; native snapshots are not a substitute for independent backup copies.

Changed Region Tracking (CRT) and deduplication reduce replication bandwidth by sending only changed extents.

Self-service restore via NGT allows users to restore individual files from VM snapshots without administrator involvement.

Recovery plans can include non-Nutanix resources and scripts, and should be tested regularly; a DR test validates RPO/RTO assumptions before a real event.