This is a machine-translated text that may contain errors!
The IT industry uses a lot of power. Data centers, servers, network equipment, PCs, mobiles, charging, cooling. All of this requires energy, and all the equipment has a limited lifespan. When we make choices about infrastructure, we are also making choices that affect the environment.
How big is the problem actually?
Some numbers to consider:
- Data centers worldwide use approximately 1-2% of the world’s total electricity consumption
- The IT industry as a whole has a carbon footprint the size of the aviation industry
- A single Google search uses approximately 0.3 Wh. It sounds small, but with billions of searches a day, it adds up.
- Training large AI models can use as much electricity as a small Norwegian house uses in a whole year
Norge er i en gunstig posisjon
Most of the electricity in Norway comes from hydropower, which makes Norwegian data centers among the greenest in the world. This is one of the reasons why international companies are building data centers in the Nordic region.
E-waste
E-waste (electronic waste) is a growing problem. Servers, PCs, mobiles and network equipment contain rare metals and toxic substances that must be handled properly.
| Problem | Consequence |
|---|---|
| Short lifespan of equipment | More waste, more production, more resource consumption |
| Toxic materials | Lead, mercury and cadmium can pollute soil and water |
| Rare metals | Extraction damages the environment and often occurs under poor working conditions |
| Lack of recycling | Only about 20% of e-waste is recycled globally |
What can you do as an IT operator?
You don’t need to save the world alone, but the choices you make in your daily work actually have an effect.
Virtualization and Consolidation
Instead of having a physical machine for each service, we use virtualization. One server with Proxmox can run 10-20 VMs or hundreds of Docker containers. This means fewer physical machines, less power consumption, and less cooling needs.
Example:
| Without virtualization | With virtualization |
|---|---|
| 5 physical servers | 1 physical server |
| 5 × power consumption | 1 × power consumption (+ a little more) |
| 5 × cooling | 1 × cooling |
| 5 × maintenance | 1 × maintenance |
Correct Sizing
An over-provisioned server (too much RAM, CPU, disk) uses more power than necessary. A VM with 16 GB of RAM that only uses 2 GB wastes resources. Start small and scale up as needed.
Extend the lifespan of equipment
Don’t replace hardware just because something newer exists. A server that works well doesn’t need to be replaced every three years. Upgrade RAM or disk instead of buying new hardware. Remove and recycle what you don’t actually need.
Turn off what you don’t use
Test VMs running 24/7 without reason use power for nothing. Use snapshots and turn off VMs when they are not in use. Set up automatic shutdown of test equipment outside working hours.
Cloud Services and On-Demand
Cloud services have the advantage that you only pay (and use electricity) for what you actually use. A VM in Azure that runs 8 hours a day uses less than one that runs 24/7. But be aware: cloud resources you forget to delete also use electricity.
Software and Resource Usage
It’s not just hardware that affects power consumption. Inefficient software uses more CPU time, which means more power.
- A website that loads 20 MB of JavaScript uses more bandwidth and processing power than one optimized to 500 KB
- A poorly written SQL query that scans the entire database uses more resources than one that is properly indexed
- A container running an entire Linux distribution for one small service uses more than one based on a minimal base image (e.g. Alpine)
This is an area where Operations and Development overlap: the developer writes efficient code, the operator ensures it runs on appropriately sized infrastructure.
Summary
- The IT industry has significant energy consumption and produces a lot of e-waste
- Virtualization reduces the number of physical machines and thus power and cooling
- Proper sizing avoids wasting resources
- Extend the lifespan of hardware instead of replacing unnecessarily
- Turn off what you don’t use, and clean up cloud resources
- Efficient software uses fewer resources than inefficient software