9 minutes
On Monday morning, the help desk receives five reports. It takes a long time to log in. A session freezes in the middle of a Teams call. An application is responding slowly. A user restarts their computer twice because that usually fixes the problem. All five reports are resolved, since all five are minor issues. The following week, they come back.
Nothing is broken in the system itself. The servers are running, the monitoring is showing green, and there’s no outage that you can pinpoint on a dashboard. Yet everyone on the work floor knows that things are running slower than they did a year ago.
What usually happens in these situations is that the environment has grown beyond the point where the original design still made sense. As more users were added, more virtual desktops were added. As more applications were added, more work was added. But the underlying capacity didn’t keep pace.
I witnessed this firsthand at a large organization where the number of VDIs increased quarter after quarter. They expanded, rolled out, and delivered everything neatly and efficiently. The only problem was that no one was looking at what CPU capacity was actually available. It wasn’t due to a lack of willingness or a lack of knowledge. There was simply no one whose job it was to ask that question.
What is the Peter Principle, and what does it have to do with your digital workplace?
The Peter Principle dates back to 1969 and is about people. The core idea: someone who performs well gets promoted one step, and then another, until they reach a position where they can no longer perform well. They then get stuck there. Yesterday’s success is precisely the reason for the step that goes too far.
In a digital workplace, the mechanism works exactly the same way, except that the “step” isn’t a promotion but an increase in workload.
- For people: performing well leads to a more demanding role—until that role becomes too demanding.
- In technology: smooth operation leads to more users, more desktops, and more applications, until the underlying hardware can no longer keep up.
- With people, this becomes apparent in performance reviews and employee turnover.
- In technology, it goes unnoticed because there is never a moment when someone formally assesses whether the design is still appropriate.
That last difference is the most important one. An employee who gets stuck will say so at some point. A system that gets stuck says nothing. It just slows down, and the rate at which that happens is so slow that users adjust their own work pace before anyone even notices.
What goes wrong if no one monitors available capacity?
Overcommitting CPU resources isn’t a mistake in and of itself. In fact, it’s precisely why virtualization is so appealing: an office worker doesn’t use their processor for the vast majority of the day. The problem arises when the ratio between what you allocate and what’s physically available quietly increases with every expansion, without anyone monitoring that ratio.
In practice, the consequences look like this:
- Slow performance is considered normal. Users tend to log in earlier, time their coffee breaks around the system startup, and eventually stop complaining.
- The diagnosis goes in the wrong direction. The laptop, the network, or the application vendor gets the blame, while the bottleneck is actually in the host layer.
- Minor incidents pile up. Each ticket is too small to warrant a project, but together they add up to a significant number of hours.
- Investments come at the wrong time. Scaling up under pressure is more expensive than scaling up based on a forecast.
- Maintenance gets pushed back. When the environment is already stretched thin, taking a host out of production for patching becomes a risk rather than a routine task.
- Trust in IT declines. That is the damage that lingers the longest.
That loss of productivity is greater than most organizations realize. A study by Markteffect commissioned by Sharp, conducted in April 2026 among 509 Dutch respondents, shows that nearly a quarter of workers lose at least fifteen minutes every day due to IT-related frustrations such as slow systems, malfunctions, and login problems. Converted, this amounts to more than 3.4 billion euros per year for the Dutch economy.
Source: allesoverhr.nl, based on Sharp/Markteffect research. https://www.allesoverhr.nl/nieuws/productiviteitsverlies-door-storingen-en-trage-systemen-onderschat-probleem-op-de-werkvloer/
Do the math for your own organization. Fifteen minutes a day, with 100 employees, adds up to well over 5,000 hours a year. That’s more than three full-time equivalents that don’t show up anywhere on a dashboard.
What actually works to prevent this?
The solution is rarely a large-scale project. It’s mainly about creating a moment when someone asks the question that no one is asking right now.
Measure wait time, not just usage
A CPU running at 60 percent load doesn’t tell you much. What you want to know is how long a virtual machine has to wait before it’s next in line. In VMware environments, this metric is called “CPU Ready,” and the rule of thumb from VMware’s architecture documentation is to keep it at 5 percent or lower. Hyper-V and other hypervisors have similar metrics. That wait time is exactly what a user experiences as lag.
Source: VMware, Architecting a vSphere Compute Platform. https://download3.vmware.com/vcat/vmw-vcloud-architecture-toolkit-spv1-webworks/Core%20Platform/Architecting%20a%20vSphere%20Compute%20Platform/Architecting%20a%20vSphere%20Compute%20Platform.1.019.html
Determine the cost of a user profile
A call center agent, a controller working with large spreadsheets, and a designer using CAD software represent three different workloads. As long as all three are assigned the same default profile, your capacity calculation will, by definition, be inaccurate. Two or three profiles are sufficient; any more than that only complicates matters.
Make capacity a part of every expansion
Adding a virtual desktop takes just a few minutes. Purchasing additional hosts requires a business case, a budget review, and a discussion that no one is eager to have. That imbalance explains almost everything. Therefore, link a quick check to every request for additional desktops: how much space is left, and when will it run out?
Schedule an annual design review
Set aside incidents, set aside your vendor, and set aside the question of which product might be added. The only question is: does this design still meet our current needs? Half a day is usually enough.
Don’t be afraid to remove things
Sometimes more capacity is the answer. Just as often, the answer is to remove something. A redundant security layer, a profile solution left behind after a migration, a package that hasn’t been opened by anyone in three years but still loads on every desktop. Cleaning up often yields better results than buying more, and it’s considerably cheaper.
What objections do I hear most often?
Our monitoring shows “green”
That’s probably true. Traditional monitoring checks availability: is it on, is there disk space, does the service respond? User experience is something else entirely. An environment can be fully available yet still take two minutes to log in. So you are measuring, just not what your users are experiencing.
Isn’t overcommitment perfectly normal?
Certainly, and I would never advise against it. The point isn’t that you overcommit; the point is that virtually no one can say what the current ratio is or when it was last checked. A ratio without an owner will always creep up.
We’re moving to the cloud anyway, so the problem will resolve itself
Partly true. In the cloud, scaling up is a matter of configuring rather than purchasing. But the capacity issue doesn’t go away—it just turns from a technical problem into a monthly bill. Anyone who migrates without clear user profiles carries over exactly the same imbalances and ends up paying for them every month.
We don’t have time for that
Understandable, and that’s exactly why it gets put on the back burner. Still, it’s a matter of spending half a day gathering data versus months of minor incidents that nobody adds up.
Wouldn’t our vendor report that?
This calls for an honest answer—including one about my own industry. Expansion generates revenue for a vendor; cleanup does not. That doesn’t mean parties are acting in bad faith, but it does explain why the question rarely comes up on its own. New Yard also sells licenses and carries out implementations. That’s precisely why I think you should dare to ask that question out loud—to us just as much as to any other party.
How do you check whether your workplace is still within its limits?
Go through these points. If you can’t answer three or more of them, that in itself is the answer.
- Do you know how many virtual desktops are currently running, and how many there were twelve months ago?
- Do you know how many physical cores correspond to that, and whether that number has changed during that period?
- Do you know the current ratio between allocated and available CPU capacity?
- Do you measure CPU wait time and not just CPU usage?
- Do you know how long it takes to log in on Monday mornings—measured, not estimated?
- Have user profiles been established, and do they still match what people actually do?
- Is there a specific time of year when the design is reviewed, independent of incidents?
- Do you know when your capacity is expected to be fully utilized, expressed in months?
- Can you take a host out of production for maintenance without users noticing?
- Is there software on the workstations that you could remove without any consequences?
How does New Yard view the digital workplace?
New Yard specializes in the digital workplace for small and medium-sized businesses and operates independently of any specific vendor. In practice, this means that I first assess what’s already in place and why, before adding a new product. In addition to consulting, we also handle licensing and implementations, so I always make it clear which role I’m fulfilling.
During a workplace health check, I assess capacity, design, management burden, and user experience, and I provide an overview that outlines what’s working well, what’s causing issues, and what I believe can be eliminated. That last part is most often overlooked, yet it usually yields the greatest benefits.
Curious to see where your workplace stands?
There’s a practical reason not to put this off. Capacity issues rarely arise at a quiet time. They tend to surface at the start of a new season, during a migration, when a new department is being onboarded, or when a host goes down and the rest have to take over. That’s exactly when there’s no time to analyze the situation, and you end up paying top dollar for rush service.
So, schedule a no-obligation introductory meeting. Together, we’ll review how your workplace is structured, what you’re currently measuring, and what data is missing. There’s no obligation, and it will definitely give you a clearer picture. Visit newyard.nl for contact information and more details.