Virtualization · High availability
Hyper-V failover cluster setup and high availability
IT Perfection designs and builds Windows Server failover clusters for Hyper-V, so virtual machines can move between hosts for maintenance and restart on another host if one fails. We size the nodes, build the storage and networks, and test failover before handover.



Is this you?
When a cluster is worth it
- A single host failure would stop the whole office
- Patching the host means taking every server down
- You have two hosts but virtual machines cannot move between them
- An existing cluster shows warnings nobody has looked into
- The cluster has never been tested with a real failure
- Storage or network for the cluster has no redundancy
What a cluster does
What high availability does and does not do
It helps to be clear about this before spending on a second host.
- Planned maintenance: virtual machines move to another node while running
- Host failure: virtual machines restart on another node after a short interruption
- It does not protect against deleted files, ransomware or a damaged database
- It does not replace backups or an off-site copy
- It does not help if the shared storage itself is lost
- It needs every node to be patched and monitored, or it fails when needed
A failover cluster restarts virtual machines automatically and needs shared storage. Hyper-V Replica copies virtual machines to a second host on its own disks, and someone starts them there if the first host fails. Replica costs less and suits two hosts or a second site. Many setups use a cluster on site and a replica elsewhere.
Each node needs Windows Server licenses that cover the virtual machines it might run after a failover, not just the ones it runs on a normal day. We include this in the design.
Scope of work
What the cluster build includes
A cluster is only as available as its weakest part: storage, network or quorum.
Availability design
We start with which virtual machines must keep running and how long an interruption is acceptable. The nodes are then sized so the remaining hosts can carry the full load when one is down, which is what makes failover work in practice.
Nodes and validation
Each node gets the same Windows Server version, updates, drivers and firmware. We run the cluster validation tests that Microsoft provides and resolve what they report before the cluster is created. Passing validation is also a condition of Microsoft support.
Shared storage
Clustered virtual machines live on storage every node can reach, such as a storage array, or the disks inside the nodes pooled with Storage Spaces Direct. We set up Cluster Shared Volumes with redundant paths so one cable or switch failure does not take the storage offline.
Cluster networks
Management, live migration, storage and virtual machine traffic are separated and given redundant ports. Live migration is secured and limited to its own network, so moving a virtual machine does not slow the others.
Quorum and witness
The quorum decides which nodes keep running if they lose contact with each other. For a two-node cluster we add a witness, either a file share or a cloud witness in Azure, so the cluster can survive the loss of one node without stopping.
Failover testing and updates
Before handover we move virtual machines between nodes, then shut a node down without warning and confirm they restart on the other. We also set up a routine for patching nodes one at a time, so updates do not need a full outage.
Project plan
How a cluster project runs
Review
We look at current hosts, storage, networks and which servers must stay available.
Design
You get the node, storage and network design, the witness choice and a written quote.
Build
Nodes, storage and networks are installed, validated and formed into a cluster.
Migrate
Virtual machines are moved in and made highly available, one at a time.
Test
We test live migration and a node failure with you, then hand over the documentation.
Every engagement starts with a written scope and a clear quote, so you know the expected cost before work starts. How pricing works
Handover pack
What you have at handover
A cluster that has been tested and can be maintained.
- A diagram of nodes, storage, cluster networks and the witness
- The validation report from the final configuration
- The results of the failover tests
- A patching routine for the nodes
- Monitoring for node, storage and quorum health
- Admin accounts in your company's name
A cluster that has never failed over is an assumption. We run the failure test with you and record how long the virtual machines took to return.
Clusters drift when one node is patched and another is not. We can monitor and maintain the cluster after handover, or pass the routine to your IT team.
Platforms and tools
Platforms we work with
Free guides from our IT library on this topic:

Who does the work
Ali Hassani, CISO
IT Perfection is led by Ali Hassani, who has 25+ years of experience in IT operations, Microsoft infrastructure, networking and cybersecurity. Design, implementation and troubleshooting work is planned and reviewed by Ali personally.








Need an independent security audit, compliance readiness or vCISO support instead? IT Perfection handles design, implementation and day-to-day IT; OC Security Audit is the related cybersecurity brand for audits, compliance readiness and vCISO work.
Larger or multi-vendor project?
We run it under IT project management: one written plan, one point of contact, vendors coordinated, and a tested rollback for every cutover.
Related work
Related virtualization services
FAQ
Hyper-V failover clustering: common questions
How many hosts does a cluster need?
At least two, plus a witness. Three or more nodes give more room for maintenance and failures. The nodes must be sized so the others can carry the load when one is down.
Does a cluster mean no downtime?
No. For planned work, virtual machines move between nodes while running. If a node fails without warning, its virtual machines restart on another node, which means a short interruption while they start up.
Do we still need backups?
Yes. A cluster protects against a host failing. It does not protect against deleted data, ransomware or corruption, because the damaged data is what every node sees. Backups with an off-site copy are still required.
What shared storage do we need?
Either a storage array that every node connects to, or Storage Spaces Direct, which uses the disks inside the nodes. The right choice depends on the number of nodes, performance needs and budget. We compare them in the design.
What is a witness?
A small extra vote that lets the cluster decide which side keeps running if the nodes lose contact. It can be a shared folder on another server or a cloud witness stored in Azure.
Can you review or fix an existing cluster?
Yes. We run validation, review the event logs, storage paths, networks and quorum settings, and give you a list of findings in order of risk. Fixes are scheduled so the cluster stays in service.
Do the hosts need to be identical?
They should be as close as possible. Different processor generations can still be clustered with a compatibility setting, at some cost in features. Matching hardware, firmware and drivers avoids most cluster problems.
What does it cost?
The price depends on the number of nodes, the storage and how many virtual machines move in. After the review you get a written scope and quote before any work starts. How pricing works.
Be ready for the day a host fails.
Start with the free 30-minute review. No obligation, and you leave with a clear list of what is fine and what needs attention.
On site across Orange County: Irvine · Newport Beach · Costa Mesa · Tustin · Santa Ana · Lake Forest · Anaheim · Fountain Valley
