Red Hat OpenShift Virtualization: FAQs from the field

Red Hat OpenShift Virtualization is an increasingly prominent topic of discussion within the virtualization community, reflecting a growing interest from traditional virtualization users. These users generally expect feature parity and frequently compare OpenShift Virtualization to their current virtualization platforms.

Discussions about OpenShift Virtualization generally follow a common pattern where users ask numerous questions after watching a high level overview. Once these questions are answered, they tend to spend more time learning about OpenShift Virtualization and often want to see it in action through demos and hands-on workshops.

Learn OpenShift Virtualization

In this article, I'll try to answer most of the common questions I've been asked by users (and potential users) in the field. Hopefully these will help set the right expectations and provide some clarity about what OpenShift Virtualization can (and can't!) do.

Is Red Hat OpenShift Virtualization a separate product or is it part of Red Hat OpenShift?

Red Hat OpenShift Virtualization is not a separate product, it is a feature of Red Hat OpenShift. This means that if you have an OpenShift subscription, you have full access to OpenShift Virtualization as it is already included in your subscription.

Do I need a full platform team to run and manage OpenShift Virtualization?

No, you still can opt to use the core OpenShift Kubernetes Engine on its own. If your focus is traditional virtualization, OpenShift Kubernetes Engine lets you explore the benefits of Red Hat OpenShift in an entry-level solution with access to the administration console, Red Hat Enterprise Linux CoreOS and OpenShift Virtualization.

Which hypervisor does OpenShift Virtualization use?

OpenShift Virtualization uses Kernel-based Virtual Machine (KVM), the Linux kernel-based hypervisor that Red Hat and most cloud providers have been using for over 15 years. This is a tier 1 hypervisor which means it runs directly in the Linux kernel and interacts with the host hardware to manage the guest operating system.

Does OpenShift Virtualization provide its own storage and network management?

Yes, VMs running on OpenShift Virtualization use the same cluster services provided to containers. They are provisioned with storage through OpenShift Container Storage Interface (CSI) plugins by creating persistent volume claims. Additionally, VMs use OpenShift software-defined networking (SDN) to establish various network interfaces relying on Pod network, while extra network interfaces can be added using Multus.

Does OpenShift Virtualization support full high availability, multi-site deployment, Regional Disaster Recovery and Metro Disaster Recovery?

Yes, these are all supported deployment models. With Red Hat technology or tested ecosystem solutions, minimum high availability (HA) can be achieved by creating a cluster with two worker nodes. Availability can be further increased by stretching the cluster across multiple availability zones. OpenShift can be installed across sites with asynchronous or synchronous data replication setup.

Where can we run OpenShift Virtualization as of OpenShift 4.15?

OpenShift Virtualization can be run on bare metal infrastructure, and it is also supported on Red Hat OpenShift on AWS (ROSA) by adding metal Amazon EC2 instances to ROSA clusters.

Does OpenShift Virtualization support application clustering?

Yes, the Red Hat Enterprise Linux (RHEL) High-Availability Add-on (Pacemaker) and Windows Server Failover Clustering (WSFC) are supported.

What operating systems are currently supported on OpenShift Virtualization?

There is a long list of supported operating systems (OSs), and you can see the full list here.

What CPU architectures is OpenShift Virtualization supported on?

OpenShift Virtualization is supported on X86 Intel and AMD based systems.

Is CPU overcommit supported?

Yes, CPU overcommit is supported on OpenShift Virtualization and can be controlled from the spec.domain.cpu (sockets, cores, threads) section.

Is memory overcommit supported?

Safe memory overcommit is available as a tech preview in the 4.16 release of Red Hat OpenShift. You can see what else is new in OpenShift Virtualization 4.16.

Does OpenShift Virtualization support GPU use?

Yes, OpenShift Virtualization supports this either by using PCI passthrough that lets you access and manage hardware devices from a VM, or by using vGPU through the NVIDIA GPU Operator.

Does OpenShift Virtualization have dual-stack network support (IPv4 and IPv6)?

Yes, but this requires your OpenShift cluster be configured with IPv4 and IPv6 dual-stack networking enabled, at which point you can control how your VMs use IPv4, IPv6, or both, by defining the spec.ipFamilyPolicy and the spec.ipFamilies fields in the Service object. OpenShift provides full flexibility on how to use IPv for single-stack or to define the order of IP families for dual-stack to meet VM deployment requirements.

Does OpenShift Virtualization support VM snapshot, live migration and backup and restore?

Yes, these are all supported.

Snapshots: A VM snapshot can be created from a running VM (hot snapshot) or a stopped VM (cold snapshot). In the case of a hot snapshot, the file system will be frozen while the snapshot is being taken and released when that is done. Once the snapshots are created, you can restore VMs from snapshots by using either the OpenShift web console or the command line.

Live migration: Moving a VM between cluster nodes requires your cluster to have shared storage with ReadWriteMany (RWX) access mode. Live migration doesn’t interrupt the VM workload and by default the live migration traffic is encrypted using Transport Layer Security (TLS).

Backup and restore: VMs can be backed up and restored using the OpenShift API for Data Protection (OADP), through either the web console or declaratively using command line as part of an automated backup and restore scenario.

Does OpenShift Virtualization provide VM observability (logging and metrics)?

Yes, OpenShift Virtualization provides observability using OpenShift logging and metrics cluster services. Metrics can be accessed through the VM web console interface where a quick dashboard is always available. Each VM also has its own tab in the interface where you can find more detailed monitoring. You can also configure a VM health check dashboard using readiness, liveness, guest agent ping probes and a VM watchdog.

Does OpenShift Virtualization provide resource scoping, map organization structure and fine-grained role-based access control?

Yes. Two levels of segregation and resource allocation can be achieved.

Logical segregation: This is achieved by relying on OpenShift multi-tenancy and mapping tenants (organizational unit) to namespaces inside a shared OpenShift cluster. This allows resources to be allocated to the tenant by adjusting the namespace quota. In other more complex scenarios, organizational units can be assigned multiple namespaces and then the quota can be controlled using multi-project quota objects.  In all cases, both access rights and network isolation can be configured using OpenShift role-based access control (RBAC), Network Policy and router sharding.

Physical segregation: Here tenants are mapped into separate OpenShift clusters, allowing resources to be controlled at the cluster level. If you have a growing number of clusters, it is recommended that you use Red Hat Advanced Cluster Management for Kubernetes as a fleet management system, giving you a single pane of glass for cross-cluster management and observability.

Note: There can be a middle ground between these options by allocating nodes within a shared cluster to organizational units and then controlling VM scheduling to always place them on the proper assigned nodes.

Does OpenShift control plane require a lot of resources on each virtualization cluster?

Not necessarily. Hosted Control Planes enable the rapid creation of tenant OpenShift clusters where the master nodes are containerized and run on the infrastructure/hub cluster. This significantly reduces the resource overhead required.

Does OpenShift Virtualization support high performance VMs and high I/O operations (IOPs) requirements?

These are both supported by using non-uniform memory access (NUMA)-aware node scheduling and storage classes.

NUMA is a compute platform architecture that allows different CPUs to access different regions of memory at different speeds. The NUMA-aware scheduling aligns the requested cluster compute resources (CPUs, memory and devices) in the same NUMA zone to process latency-sensitive or high-performance workloads more efficiently. It also improves pod density per compute node for greater resource efficiency.

Storage classes can be created for different CSI plugins allowing cluster administrators to provide different classes that reflect the performance of the underlying storage provider and enable persistent volumes to be dynamically provisioned.

Does OpenShift Virtualization support command line interface (CLI) and DevOps-based VM provisioning?

Yes. VMs are managed as custom resource definitions (CRDs) in OpenShift, allowing administrators to declaratively create and manage them using the OpenShift CLI and YAML manifests. Administrators can also use the virtctl CLI utility to simplify creating manifests for VMs, VM instance types and VM preferences. Finally, VM lifecycle management can be also integrated inside OpenShift pipelines for better automation using the SSP operator and its Tekton supported tasks. The operator provides Tekton tasks to help with VM management including: create-vm-from-manifest, create-vm-from-template, copy-template and modify-vm-template.

Does OpenShift Virtualization provide some level of VM resiliency?

Yes, but at different levels, including:

• Self-heal
• Manual or automatic rescheduling VMs on healthy nodes in case of cluster node failure
• Snapshots that can be restored, in addition to backup/restore facilities
• Support for Regional and Metro DR

Does Red Hat OpenShift Virtualization provide a VM catalog and self-service portal?

Yes, through VM templates and a cloud-like virtual instance catalog. Templates can also be customized and provided to individual users if required.

Does Red Hat OpenShift Virtualization support migration from other virtualization platforms?

Yes, including VMWare, OpenStack, Red Hat OpenShift Virtualization, and Red Hat Virtualization.

Does OpenShift Virtualization have cross-cluster management capabilities?

Yes, Red Hat Advanced Cluster Management for Kubernetes is used for fleet management allowing administrators to do multi-cluster management, cross-cluster observability and cross-cluster resource searching.

We have a huge number of VMs, is there any way we can speed up our migration?

If you are planning to migrate hundreds or thousands of VMs to OpenShift Virtualization, Red Hat Ansible Automation Platform combined with the Red Hat Migration Toolkit for Virtualization can help you build a migration factory and speed up the migration process. Watch this 3-minute demo that shows how it works.

Are there other quick demos that show some of OpenShift Virtualization features?

There are lots of OpenShift Virtualization demo videos out there. Here are 9 to help you get started:

• VM provisioning and lifecycle
• Declarative VMs - CLI and web console
• Catalog and cloud like experience
• Custom template
• Network Management
• VM Snapshots
• VM Backup and Restore
• Windows VM
• VM Provisioning, Cloning, and Load Balancing

Curious to see how OpenShift Virtualization can work for your business?

Learn from Red Hat experts and get hands-on experience with virtualization tasks at the Red Hat OpenShift Virtualization Road Show.