DATA SHEET AT-02-V · VIRTUAL SERVERSREV. 2026-A

Virtualisation: several servers on one machine

Virtualisation means that several separate Windows servers, each for its own task, run on one powerful machine. Instead of three ageing boxes in the back room there is one, and at the next hardware replacement the servers move across as a whole, without a reinstall.

We plan and support such environments with Hyper-V or VMware for companies in the Vogtland. That includes the uncomfortable side: if the one machine fails, all the servers on it are down. We settle how you carry on working in that case before the move. You can reach us Mon–Fri 8:00–17:00 on +49 3744 4399760; the initial consultation is free.

Discuss your project

When virtualisation is worth it for your business.

The typical case: in the back room there are two or three servers, each a few years old, each for exactly one task. One holds the files, one the inventory and order system, one the sign-in. Each needs power, space, spare parts and, one day, a successor. With virtualisation the same servers run as separate virtual machines on a single current machine.

What you get from it: less hardware that can age, and a backup that captures whole servers at once. The next hardware replacement can be planned, because the virtual servers move across to the new machine as a whole. And if a second specialist application comes along, it does not need a box of its own, just another virtual server.

It is not always worth it. If you have one server with one task and nothing more is planned, a single solid server can be the more honest solution. We will tell you so. Two questions decide it, and they have nothing to do with technology: how long can your business stand still, and how much freshly entered work could be lost in the worst case?

This page goes into planning and the move in depth. You will find the overview under Servers & virtualisation. What a Windows server does in operation and how we set up users, permissions and shares is on the server page. We never look at servers in isolation, but as part of your ongoing IT support.

One machine, several servers. And a shared risk.

Three terms are enough. The host is the physical machine standing in the cabinet. On it runs the hypervisor (Hyper-V from Microsoft or ESXi from VMware), a slim piece of software that divides up the host’s processor, memory and drives. And inside it run the virtual machines, VMs for short: complete Windows servers that behave as if they had hardware of their own.

For your staff nothing changes. The drive has the same name as before, the inventory and order system starts as before. The servers are cleanly separated from one another: if the application server hangs, the file server keeps running, and restarting one does not disturb the other.

The dashed lines in the figure show the flip side. All the VMs depend on the same host. If it fails (power supply, mainboard, controller), all the servers on it are down at the same moment. Before, perhaps only one of three would have failed. That is why we do not skimp on the host, and why there is a section of its own further down for the day when it goes down anyway.

ONE HOST: SHARED FAILURE ZONEHARDWARE: CPU, RAM,STORAGEHYPERVISOR: HYPER-VOR VMWAREVIRTUAL MACHINESON IT: VMS WITH THEIR OWN GUEST SYSTEMVM: FILESVM: APPLICATIONVM:DIRECTORY SERVICEONE HOST: SHAREDFAILURE ZONEHARDWARE: CPU, RAM,STORAGEHYPERVISOR: HYPER-VOR VMWAREVIRTUAL MACHINESON IT: VMS WITH THEIR OWNGUEST SYSTEMVM: FILESVM: APPLICATIONVM: DIRECTORY SERVICE
FIG. 1 · SEVERAL VMS ON ONE HOSTDASHED = SHARED

What has to come along, and what depends on what?

Before the move we take stock, on site, of what is running today: which servers, which programs, which data, and who talks to whom. You do not have to gather any documents for this. If there is documentation, it helps. If there is none, we write it.

The dependencies matter. An inventory and order system rarely consists of a single program: it includes a database, a folder of receipts, the label printer, perhaps a link to the web shop. If one of them does not move with it or comes back up in the wrong order, the whole thing grinds to a halt.

The order is also why we look at the sign-in first. Active Directory (the central user and permissions management) and DNS (the internal address book) have to be running before any other server can start sensibly. After a power cut or a restore, the virtual servers therefore start up in a fixed order, and that order is in the system file (Systemakte).

What we cannot find out stays on the list as an open item, visible to you. We do not build guesses into a quote.

Hyper-V or VMware? Your existing set-up decides.

Both platforms can do what a small or medium-sized business needs, and we are not tied to either vendor. The decision comes down to three things: what you already have and know, what the vendors of your software approve, and what the licences cost.

Approval is where things often get stuck. Some industry software is certified only for a particular combination of Windows version, database and platform. We ask the vendor before we order anything. Without approval you get no help from them if something goes wrong.

Licence dongles (the small USB plugs without which some CAD or industry software will not start) and devices attached directly to the server are a subject of their own. A plug sits in one physical machine, yet in an emergency the virtual machine is supposed to be able to move to another. There are solutions for this, you just have to plan them in beforehand. We clarify what your vendor supports.

And the licences. Windows Server is licensed by the processor cores of the host, and depending on the edition, a different number of virtual Windows servers may run on it. On top of that come the client access licences for your users or devices. The rules are in Microsoft’s licensing guidance on virtualisation (in English). We work it out for your case, and the quote (Angebot) states which licences you already have and which have to be added.

How many servers fit on one host?

Processor: what is working at the same time?

The virtual servers share the host’s processors. That works well as long as they do not all want to compute at the same time. So we look at when your peaks fall (the sign-in in the morning, the end-of-day close in the evening) and take measurements on your existing servers. Simply giving a sluggish inventory and order system more virtual processors often does not help, by the way. If the problem lies with the drives, it will not get faster.

Memory: skimping does not pay here.

Every virtual server needs its own memory, and the host needs some on top. We add up what your servers actually use today, and for industry software we take the vendor’s specification. Whether a server gets a fixed amount of memory or only as much as it currently needs, we decide per application. Databases are sensitive to this.

Drives: space is one thing, speed another.

The hard disks of all the virtual servers sit on the host’s storage as large files. If several databases write at once and the backup runs alongside, fast drives matter, not many terabytes. And the space needs headroom: if this storage fills up, all the virtual servers stop at once. That is why we monitor the free space and get in touch before it gets tight.

Then there is the reserve, and it has a defined purpose. With a single host it is room for growth and for the moment when backup and daily business coincide. With two hosts, one must be able to carry all the important servers alone, otherwise the second is no use in an emergency. The quote (Angebot) shows the figures we worked with: the virtual servers’ needs, the host’s own needs, reserve.

ADD UP PER RESOURCEVM NEEDSHOST NEEDSOPERATING RESERVEPLANNEDCAPACITYFOR EACH RESOURCE SEPARATELYCPU: LOAD OF THE VMSRAM: MEMORY OF THEVMSSTORAGE I/O:ACCESS PATTERNSADD UP PER RESOURCEVM NEEDSHOST NEEDSOPERATING RESERVEPLANNED CAPACITYFOR EACH RESOURCE SEPARATELYCPU: LOAD OF THE VMSRAM: MEMORY OF THE VMSSTORAGE I/O:ACCESS PATTERNS
FIG. 2 · NEEDS AND OPERATING RESERVE

What happens if the host fails?

This is the question that has to be answered before the purchase. There are essentially two answers. They differ in price and in the time your business stands still.

Answer one: one host and a rehearsed restore. If the host fails, we repair or replace it and restore the virtual servers from the backup. Until then the business is at a standstill. You know beforehand how long that takes, because we have run through it once. If such a standstill is annoying but bearable for you, this is the right solution.

Answer two: two hosts in a group, the cluster. If one fails, the other restarts its virtual servers. That happens by itself, but not without interruption: the servers boot up again, the database checks itself, and your people sign in again afterwards. In VMware the function is called vSphere HA; Broadcom describes which virtual machines are restarted in the process (in English). In Hyper-V the failover cluster takes care of it.

For two hosts to agree which of them is still running, the group needs a third vote (in technical jargon: quorum with witness). Without it, if the connection between them fails, each assumes it is the last one standing, or both stop to be on the safe side. We set that up as well.

Two hosts only help if they do not share a single point of failure. If both are plugged into the same switch and the same power strip and access the same single storage, you pay twice for no extra safety. So we record what the two have in common and separate what can be separated with reasonable effort (on the network side, that overlaps with our networks service). And the remaining host must be strong enough to carry all the important servers alone.

There is one thing neither answer covers: a database that has corrupted itself, and the deleted file. The second host restarts a broken database just as broken. For that you need the backup.

How much work may be lost, and how long may it take?

For each important application we set two values with you. The first: how much work may be gone in the worst case, an hour or a day? (Specialists call this RPO.) The second: how long may it take until people can work again? (That is the RTO.) The first determines how often backups are made. The second, whether one host is enough.

Three things are often mistaken for a backup. The snapshot (in Hyper-V it is called a checkpoint) freezes the state of a virtual server, for instance before an update, so that you can go back quickly. But it lies on the same storage as the server itself. If the storage is gone, the snapshot is gone with it. Broadcom says so explicitly for VMware: snapshots are not a backup (in English). We use them for a single change and delete them afterwards, because old snapshots eat space and make the server slow.

Replication keeps a continuously updated copy of a virtual server ready on a second machine. Handy for a quick restart. But it copies every mistake too: the deleted file is deleted in the copy shortly afterwards. And RAID, the disk array in the host, catches a failed disk and nothing else.

Only something that is stored independently of the host, keeps older versions and can demonstrably be restored counts as a backup. That the backup job starts every night is not enough; what matters is whether it completes and whether the database in it is usable. How we build backups is covered under Backup and disaster recovery.

CLUSTER: RESTART WITH INTERRUPTIONHOST A FAILSPREREQUISITESCHECKRESTART ON HOST BBACKUP: TESTED RESTOREINDEPENDENT BACKUPISOLATEDRESTORE TESTCHECK APPLICATION ANDDATA STATECLUSTER: RESTART WITHINTERRUPTIONHOST A FAILSCHECK PREREQUISITESRESTART ON HOST BBACKUP: TESTEDRESTOREINDEPENDENT BACKUPISOLATED RESTORE TESTCHECK APPLICATION ANDDATA STATE
FIG. 3 · AVAILABILITY AND RESTOREDASHED = ONLY IF CONDITIONS ARE MET · GOLD = TESTED RESTORE

The server starting is only half the rehearsal.

In the rehearsal we bring a virtual server back from the backup, isolated from the live network. Otherwise you would suddenly have two servers with the same name in the building, and that causes trouble.

Then comes the part that matters. We can see for ourselves that Windows boots. Whether you can actually work with it is something you see: someone from the office signs in, opens a file, calls up the latest order in the inventory and order system. Is the data up to date, is the date right, is anything missing? We agree with you beforehand what exactly is checked.

Two kinds of server need particular care here. The domain controller (the server with the user management) does not cope well with being rolled back to an old state while others of its kind are still running; Microsoft has its own procedures for it, which we follow. And databases have to be backed up in such a way that they come back consistent. We clarify both before the rehearsal and not during it.

At the end there is a report: which version, how long it took, what worked and what did not. It applies to the day of the rehearsal. If something bigger changes at your company (new software, new host), we rehearse again. What such a test shows is explained in the guide to the restore test.

From the physical box into the virtual machine: how we prepare the move.

  1. Write down what moves and what has to run afterwards.

    For each server we record what task it has, what it depends on and who at your company confirms after the move that work is possible again. That is usually not the management but the person who works in the inventory and order system every day. This list later becomes our test record.

  2. Move or set up afresh?

    We transfer a healthy server as a whole into a virtual machine. An old one running a Windows that no longer gets updates is better set up afresh, bringing over only the data; otherwise the legacy problems move too. Which way is right for which server is in the quote (Angebot), with reasons. If the vendor of your software wants or has to move its part itself, we plan that in.

  3. Trial run and date.

    Where possible we do a trial move first: the copy runs isolated on the new host, and your people check whether everything is there. That shows what is missing (a dongle, a vendor approval, a forgotten printer) while it is not yet holding anyone up. After that we set the date for the real switchover, outside your busy periods.

The plan also covers who is reachable on the cut-over day and who decides whether we carry on or abort. If we need the vendor of your software, we involve them beforehand. Nothing goes live without the OK of your named person, even if everything has technically been copied.

The cut-over day: switch, check, approve. And the way back.

On the cut-over day it runs in five steps, as in the figure. Trial and acceptance are already done. Then comes the write freeze: from a fixed point in time nobody changes data on the old servers any more, not even a background service or a web shop import. We transfer the latest data, start the new virtual servers and check them against the list from the preparation.

The approval comes from you, not from us. Your named person goes through their work steps and says: fine. Only then does everyone else sign in again. The old servers stay off, so that two versions of the data do not keep growing side by side. Your people know beforehand what changes for them that day (usually nothing but a break).

The way back hangs on a single question: has anyone worked on the new system yet? Before the approval the answer is simple. We switch the new servers off, the old ones back on, and everything is as it was the day before. We also check whether a background service on the new system has already written anything before the first person signed in.

After the approval it is different. Every newly entered order exists only on the new system. Anyone who simply switches the old server on now has lost those orders. Then we first decide which system counts, block write access to the other and reconcile the data, depending on the software together with its vendor. Under what conditions we abort and how the reconciliation runs is in the plan beforehand. The old servers also get a fixed date until which they are kept.

PLANNED SWITCHOVERTEST ANDACCEPTANCEWRITE FREEZETRANSFERFINAL DATANEW INSTANCECHECKAPPROVALCASE: WAY BACKAFTER NEWWRITESWRITE SOURCEDETERMINERECONCILE DATA,CHECK WAY BACKPLANNED SWITCHOVERTEST AND ACCEPTANCEWRITE FREEZETRANSFER FINAL DATACHECK NEW INSTANCEAPPROVALCASE: WAY BACKAFTER NEWWRITESDETERMINE WRITE SOURCERECONCILE DATA,CHECK WAY BACK
FIG. 4 · SWITCHOVER AND TESTED WAY BACKDASHED = WAY BACK PLANNED IN ADVANCE · GOLD = YOUR APPROVAL

Host and virtual servers both need care.

In a virtual environment there are two levels that need updates: the host with its hypervisor and every single virtual server. An update on the host affects all the servers on it, because it has to restart for that. So we put it in a time window that we agree with you, and afterwards check that every server is back and doing its work: sign-in, drives, inventory and order system.

In between we monitor what tends to go wrong in a virtual environment. The shared storage fills up, a forgotten snapshot grows, the backup did not complete overnight. For each of these alerts it is settled who receives it and what happens then. We handle most of it by remote access.

Everything we set up or change goes into the system file (Systemakte): structure, settings, start order, access data. It stays with you. You could change service provider without the next one having to guess.

You can reach us Mon–Fri 8:00–17:00. The response time that applies is in the support contract. If a fault lies in the industry software itself, we bring in its vendor.

What you have in hand before we start.

It starts with a conversation, by phone or at your premises. For business customers it costs nothing. Tell us what you have and why you are thinking about it: the hardware is old, new software is coming, or you finally want to sort out the back room.

After that you receive a written quote (Angebot) before anything begins. It lists survey, host, licences, set-up, move and the later support as individual items. So you can also commission only the planning and decide afterwards. What can be reused of your existing hardware is also stated there. You will find the hourly rates in the price list; for hardware orders we agree a deposit.

How much work it will be depends above all on how well your current environment is known. With documentation and passwords it goes faster. It works without them too; the project then simply starts with the survey.

Specification

DATA SHEET AT-02-V · VIRTUAL SERVERS

Platform
HYPER-V OR VMWARE · ACCORDING TO EXISTING SET-UP AND SOFTWARE APPROVAL
Guests
WINDOWS SERVER · FILES · APPLICATION · ACTIVE DIRECTORY
Sizing
PROCESSOR · MEMORY · DRIVES · RESERVE WITH A PURPOSE
Failure plan
ONE HOST WITH TESTED RESTORE OR TWO HOSTS
Backup
INDEPENDENT OF THE HOST · RESTORE TESTED
Migration
TRIAL RUN · OUTSIDE BUSY PERIODS · WAY BACK SET IN ADVANCE
Support
UPDATES FOR HOST AND GUESTS · MONITORING · REMOTE ACCESS
Documentation
SYSTEM DOCUMENTATION · ACCESS STAYS WITH YOU
Quote (Angebot)
IN WRITING BEFORE WE START · INITIAL CONSULTATION FREE
Available
MON–FRI 8:00–17:00

Prices

€90 per hour plus VAT (MwSt.). That is the standard hourly rate; for some work, for example on servers or networks, a higher rate may apply, and it is stated in the written quote (Angebot). We charge for the time we spend, rounded up to the next quarter of an hour. Travel from €20 plus VAT. If you prefer, a monthly flat rate instead of hourly billing. The initial consultation is free, by phone or at your premises; we do not charge for the travel either.

Questions & answers

Common questions, answered briefly.

What is a virtual server, in one sentence?

A complete Windows server that shares a machine’s hardware with other virtual servers and still runs separately from them. To your staff it looks like any other server.

We have three old servers. Will they fit on one machine?

Often, yes. We size the machine by measuring your three: what they actually need in processor, memory and drives, above all at the same time. There is no fixed number of servers per host. A file server is modest; a database is not.

What happens if that one machine fails?

Then all the servers on it are down until the host is repaired or replaced and the backup has been restored. We rehearse how long that takes beforehand, so you know how long it will be. If it is too long for you, a second host is the answer.

With two hosts, does everything keep running without interruption?

No. If one host fails, the second restarts its servers. That happens by itself, but it takes time until Windows and the database are back up, and your people have to sign in again. The second host is no help at all against a damaged database or a deleted file. That is what the backup is for.

Can we keep using our existing hardware as the host?

Sometimes. We check whether the platform vendor still supports the machine, whether memory and drives are enough for all the servers, and how long spare parts will still be available. A host that carries several servers should not be the oldest machine in the building.

Which licences do we need for virtual Windows servers?

Windows Server is licensed by the processor cores of the host; depending on the edition, a different number of virtual Windows servers may run on it. On top of that come client access licences for users or devices and, if you choose VMware, its own licence. We check what you already own. The quote (Angebot) shows what has to be added.

Our software needs a USB dongle. Does that work virtually?

Often yes, just not automatically. The dongle has to be passed through to the virtual machine, and that ties the machine to the device the dongle is plugged into. We check with the vendor which method they support before we commit to the move.

We take a snapshot before every update. Is that enough as a backup?

No. The snapshot sits on the same storage as the server; if that fails, both are gone. It is useful for the period right after the update and should be deleted afterwards.

Does the move bring our old Windows up to date?

No. A Windows Server that no longer receives updates will not receive them as a virtual machine either. The move is a good occasion to renew it at the same time, though. We check with the vendor whether your software can handle that.

Can we go back if something does not work after the move?

Yes, and how is settled beforehand. As long as nobody has worked on the new system yet, we simply switch the old servers back on. After that it only works if the data is reconciled, because otherwise the newly entered orders would be missing. That is why we check thoroughly before you give your approval.

What does it cost to virtualise our servers?

It is made up of host, licences and labour, and the labour depends on how many servers move and how well they are documented. After the free initial consultation you receive a written quote (Angebot) with individual items. For hardware orders we agree a deposit.

What don’t you do when it comes to virtualisation?

Proxmox and Linux servers. We work with Hyper-V and VMware and with Windows Server as the guest, because that is what we know in depth. Containers (Docker and the like) are a different subject and are not part of it either. We do not promise uninterrupted operation: you learn beforehand how long an outage would last, and the aim is that it stays short.

Contact

Are you due to replace your servers?

Tell us what is running at your company today and what you are planning. For business customers the initial consultation is free, by phone or at your premises. After that you receive a written quote (Angebot).

+49 3744 4399760Mon–Fri 8:00–17:00
CallSend an enquiry

PARTS LIST · SHEET 1/1

What this page is made of

Values measured at build time. With JavaScript, your browser measures them again live.

ITEMPARTTYPESIZE
01index.htmlDocument≈ 29 kB
02styles.cssStylesheet≈ 20 kB
03scripts.jsScript≈ 13 kB
04archivo-variable.woff2Font≈ 58 kB
05ibm-plex-mono.woff2Font≈ 30 kB
06favicon.svgImage< 1 kB

Advertising or analytics cookies: none. Without your consent, the page loads nothing from third-party servers apart from Cloudflare, our delivery service. Fonts, scripts and images are hosted by us. That is how we build your IT, too.

OUTSIDE OFFICE HOURS

Feel free to call anyway

Our regular office hours are Monday to Friday, 8:00–17:00.

In the evening and at weekends the number is forwarded to us, and often someone will pick up. We cannot promise that, though.

If nobody answers, leave a short message on the voicemail: your name, your number and what it is about. We will call back as soon as we can, on the next working day at the latest.

The call costs nothing. If someone has to start work straight away outside office hours, the surcharge in the price list applies.

Call now+49 3744 4399760

Prefer to write? Use the contact form or e-mail [email protected]. We will read it on the next working day.

DIAGNOSIS · AT-K2

What do you need help with?

↵ SELECTESC CLOSECALL: +49 3744 4399760

Custom settings

Which services may load?

Strictly necessary

The choice you make here and your settings for colour scheme and motion. These are kept as small entries on your device and are never sent to a server. They are needed for the site to work and therefore cannot be switched off.