Zombie Servers: How to Find and Eliminate Hidden Data Centre Power Waste 🧟♂️
Sustainability | Energy Management | Intelligent PDUs I A server can be powered on without doing useful work.
Across data centres, servers and other IT equipment can remain operational long after the application, project or workload they once supported has disappeared. They continue to consume electricity, generate heat and occupy valuable rack and power capacity.
These forgotten or severely underutilised machines are often referred to as zombie servers.
With electricity demand from data centres increasing rapidly, identifying this hidden consumption is becoming increasingly important. The International Energy Agency (IEA) projects global data centre electricity consumption to reach around 945 TWh by 2030, more than double current levels, with AI a major driver of that growth.
Finding zombie servers is therefore no longer just about sustainability.
It is about understanding where your power is going, whether that consumption is creating value and how existing data centre capacity can be used more effectively.
Sustainability
What is a zombie server?
A zombie server is a physical server that continues to consume electricity while performing little or no useful computational work.
A server may become a zombie after an application has been migrated, a project has ended, a customer environment has been decommissioned or responsibility for the equipment has become unclear.
The workload disappears.
The power consumption does not.
Even when it is doing little useful work, the server can continue to consume:
electricity;
cooling capacity;
rack space;
available rack power;
monitoring resources;
maintenance time.
One forgotten server may seem insignificant.
Across hundreds or thousands of racks, however, unnecessary consumption can become a much larger operational and energy management problem.
Why are zombie servers becoming more important?
Data centres are being asked to deliver more compute while power is becoming an increasingly valuable resource.
AI, cloud services, high-performance computing and continued digitalisation are increasing demand for data centre infrastructure. According to the IEA, global data centre electricity consumption is projected to grow by around 15% per year between 2024 and 2030 in its base case.
This changes the conversation around efficiency.
A server consuming power without delivering useful work is not simply wasting electricity. It may also be occupying electrical, cooling and physical capacity that could otherwise support productive IT equipment.
Finding and eliminating unnecessary loads can therefore support both:
energy efficiency and capacity optimisation.
How do zombie servers occur?
Zombie servers rarely appear because of one major technical failure.
More often, they are the result of everyday operational changes.
An application is migrated to another environment, but the original server remains powered.
A temporary project ends, but its infrastructure is never fully decommissioned.
Equipment is retained as a precaution, then forgotten.
Ownership changes between teams.
Or nobody wants to switch a server off because nobody is completely certain what will happen.
Leaving the equipment powered can appear to be the safest option.
Over time, these individual decisions can leave unnecessary loads hidden throughout the data centre.
Can you identify a zombie server from its power consumption?
No. Power consumption alone cannot tell you whether a server is a zombie server.
This distinction is important.
Imagine two servers:
Server A Power consumption: 145 W IT utilisation: 72% Status: Active
Server B Power consumption: 138 W IT utilisation: 1% over an extended period Status: Possible zombie server
Server B does not become suspicious because 138 W is inherently too high.
It becomes suspicious because it continues to consume power while doing almost no useful work.
This is why effective zombie server detection requires more than energy monitoring alone.
How do you identify zombie servers?
A strong approach is to combine physical power data with IT and operational data.
Power data answers:
How much electricity is this device actually consuming?
IT and operational data help answer:
What is this device actually doing?
Relevant information can include:
outlet-level power consumption;
CPU or workload utilisation;
historical utilisation;
application activity;
asset ownership;
DCIM information;
server monitoring;
commissioning and decommissioning records.
When these data sources are combined, potentially redundant equipment becomes much easier to investigate.
This is where intelligent rack PDUs can provide an important part of the picture.
Step 1: Make rack power consumption visible
You cannot optimise power consumption you cannot see.
At rack level, an intelligent PDU can provide information about electrical load and energy consumption.
Schleifenbauer PDU 5.0 supports measurements including current, voltage, active power and cumulative energy consumption, with measurement capabilities available at input, branch and outlet level depending on the PDU configuration.
Rack-level information helps answer questions such as:
How much power is this rack consuming?
How is the load distributed?
How much capacity remains?
But if the objective is to investigate individual servers, rack-level consumption alone is not enough.
You need to get closer to the connected device.
Step 2: Measure power consumption at outlet level
Imagine a rack containing 30 servers.
Knowing that the rack consumes 6 kW is useful for capacity management, but it does not tell you which individual devices are responsible for that load.
Outlet-level power monitoring provides that additional layer of visibility.
A Monitored PDU can measure energy consumption at both input and individual outlet level, making it possible to see how much energy connected equipment is consuming.
This can help reveal equipment worth investigating.
For example:
Outlet 12 → Server 12 → 138 W
The PDU has now established something useful:
Server 12 is consuming power.
It has not established whether Server 12 is a zombie server.
That requires another piece of information.
Step 3: Combine PDU data with IT utilisation data
Now imagine IT monitoring shows:
Server 12 → average utilisation: 1% over the past 30 days
Combined with the outlet-level measurement:
Server 12 → 138 W → 1% utilisation
This is far more meaningful.
The server is demonstrably consuming electricity while apparently performing very little useful work.
It can now be flagged as a possible zombie server and investigated by the responsible team.
This distinction matters because a low-utilisation server may still be business-critical. It could provide redundancy, run an infrequent but important application or be deliberately kept available for resilience.
The final decision should therefore remain operational.
Power data identifies the consumption. IT data provides context. Together, they provide evidence for action.
Monitored PDU: See where the power goes
For data centres that need visibility beyond total rack consumption, a Monitored PDUadds individual outlet-level energy metering.
This allows operators to understand the consumption of connected servers and other IT equipment rather than viewing the rack as a single electrical load.
Schleifenbauer Monitored PDUs combine input-level and individual outlet-level metering and are designed for applications where detailed energy visibility is required without remote outlet switching.
Finding a server with continuous power consumption and extremely low utilisation is not permission to immediately disconnect it.
Before taking action, operators should verify questions such as:
Who owns this server?
Which applications depend on it?
When was it last meaningfully used?
Is it part of a redundancy or disaster recovery strategy?
Can its workload be consolidated elsewhere?
Can it safely be decommissioned?
This creates a controlled process rather than simply hunting for low-utilisation equipment.
Once the server has been confirmed as unnecessary, however, there is another opportunity.
Remote power control.
Step 5: Switch unnecessary equipment off
Physically visiting a rack simply to switch equipment off or power-cycle a device is inefficient, particularly in large, distributed or remote environments.
ASwitched PDU allows individual outlets to be controlled remotely.
Remote outlet switching can be useful for:
controlled shutdowns;
remote power cycling;
troubleshooting;
commissioning;
maintenance;
remote locations.
But there is an important distinction between switching and monitoring.
A Switched PDU provides remote outlet control, while a Monitored PDU provides outlet-level energy metering. Schleifenbauer’s Managed PDU combines both capabilities.
Zombie servers are usually discussed in the context of sustainability.
But there is another reason to find them:
capacity.
Imagine a rack approaching its available power limit.
One option is to look for additional capacity.
Another is to first ask:
How much of the power we already have is supporting useful IT?
If unnecessary servers can be safely decommissioned, operators may recover:
rack power capacity;
cooling capacity;
physical rack space;
network resources;
operational capacity.
That capacity may then become available for productive equipment.
With AI and high-density computing increasing pressure on data centre power infrastructure, recovering existing capacity can be just as important as adding new capacity.
From individual measurements to data centre energy management
Outlet-level measurements become even more valuable when they are not viewed in isolation.
At scale, power data can be incorporated into broader monitoring and management environments.
Depending on the chosen Schleifenbauer architecture and configuration, PDU data can be accessed and integrated with management environments. The PDU 5.0 Controller architecture supports integration with external DCIM systems, while the Gateway architecture can centralise intelligence through the EnerTree Platform.
This creates opportunities to combine power information with other operational data.
Instead of asking once a year:
“Do we have zombie servers?”
the organisation can move towards continuous visibility of how rack power is being used.
What role does an intelligent PDU play in detecting zombie servers?
An intelligent PDU provides physical power consumption data.
It does not independently know whether the workload consuming that electricity is useful.
That distinction makes the role of the PDU clearer:
The PDU measures what is consuming power. IT monitoring shows what the equipment is doing. Operational context determines whether the equipment is still needed.
When those three elements come together, potential zombie servers become much easier to identify and address.
Which PDU do you need?
The right PDU depends on the level of visibility and control required.
Choose a Managed PDU when you need both individual outlet-level monitoring and remote switching.
For a structured zombie server strategy, the Managed PDU provides the most complete combination of visibility and control.
Frequently asked questions about zombie servers
What is a zombie server?
A zombie server is a physical server that remains powered on while performing little or no useful computational work. It may continue consuming electricity, cooling and rack capacity even though the workload it originally supported is no longer required.
How can you identify a zombie server?
Zombie servers can be investigated by combining physical power consumption with IT utilisation and operational information. Outlet-level PDU monitoring shows whether a server is consuming electricity, while server monitoring, DCIM or other IT data can indicate whether it is performing useful work.
Can a PDU detect a zombie server?
Not by itself. An intelligent PDU can detect and measure the electrical consumption of connected equipment. Determining whether that equipment is a zombie server requires additional context, such as workload utilisation, application activity and asset ownership.
Does high power consumption mean a server is a zombie server?
No. A server consuming 200 W could be performing useful work, while a server consuming 100 W could be almost completely idle. Zombie server identification is based on the relationship between continued energy consumption and a lack of useful workload, not on a specific wattage threshold.
What is outlet-level power monitoring?
Outlet-level power monitoring measures electrical consumption at individual PDU outlets. This provides more granular information than measuring only the total power consumption of a rack and can help operators understand how much power individual connected devices consume.
How can a Managed PDU help with zombie servers?
A Managed PDU combines outlet-level power monitoring with remote outlet switching. It can provide the physical energy data required to investigate potentially unused equipment and, once equipment has been verified as unnecessary and safely decommissioned, provide remote control of the relevant outlet.
Why do zombie servers matter for data centre capacity?
Zombie servers can consume power, cooling and rack space without delivering sufficient useful compute. Removing genuinely unnecessary equipment can therefore potentially free existing infrastructure capacity for new workloads.
Stop feeding the zombies 🧟♂️
The challenge with zombie servers is not simply that they consume electricity.
It is that their consumption can remain hidden amongst thousands of legitimate loads.
Finding them starts with visibility.
Outlet-level power monitoring can show where electricity is actually being consumed. IT and operational data can then determine whether that consumption is justified.
Once unnecessary equipment has been identified and verified, intelligent power management provides the tools to take action.
With intelligent rack PDUs, outlet-level energy monitoring and remote power control, Schleifenbauer helps data centre operators move from:
Power distribution → Power visibility → Power intelligence → Action
For more information about Schleifenbauer Products BV and our energy management solutions, visit our website: www.schleifenbauer.eu. Feel free to contact our team at Telephone +31 73 52 30 256 or info@schleifenbauer.eu.
To provide the best experiences, we use technologies like cookies to store and/or access device information. Consenting to these technologies will allow us to process data such as browsing behavior or unique IDs on this site. Not consenting or withdrawing consent, may adversely affect certain features and functions.
Functional
Always active
The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication over an electronic communications network.
Preferences
The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user.
Statistics
The technical storage or access that is used exclusively for statistical purposes.The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
Marketing
The technical storage or access is required to create user profiles to send advertising, or to track the user on a website or across several websites for similar marketing purposes.
