Get know factor, which affects the price of electricity in DC

What affects the price of electricity in a data center?

Across entire Europe and not only there, the bad habit of rising electricity prices is spreading. How relevant the increase in energy prices is, can be debated, but the truth is, that high prices are already being tackled at national level. The main cost item of any data center is energy. But what is in the price of kWh in a data center included?

Servers are placed in the data center primarily for reasons of availability and, of course, security. Security is not very dependent on the price of electricity, although every modern datacenter has an electronic access system, camera system, EPS and others.

However, electricity is crucial for availability. The redundant routers and switches that ensure continuous availability of data services won't light up without power, as well as the client devices.

The price of electricity, especially in our data center, is a complex of items including:

  • 1) Purchase of raw electricity and its distribution
  • 2) Backup system of electricity
  • 3) Cooling
  • 4) Maintenance and equipment amortization

And all this in proportion and with regard to the actual consumption of the customer's equipment in the data center.

1) Purchase of raw electricity and its distribution

Everyone knows the purchase of raw energy from the distribution network from their bills. As same as connection of the power supply lines to the equipment, and we all know this item as "distribution".

2) Backup system of electricity

A power supply in a proper data center must be backed up by UPS and diesel generators. UPS systems work with a factor of efficiency conversion and perhaps every data center wants to have the most efficient UPS circuit system in proper redundancy. Our data center operates three independent central UPS circuits with up to 98% efficiency. We cover long term power fluctuations with diesel generators, but unfortunately the price of diesel has also risen significantly, and the data center must be able to cover backup from diesel generators for at least 12 hours. Only in 2022, the electricity distribution network in Prague experienced 3 huge outages. Both of our diesel generators were always fully ready and customers did not experience any reduction in service availability. However, in one hour of back-up operation, the diesel generators consume an average of 85l of diesel and sustaining outages of up to 12 hours without refuelling is not a problem. Unfortunately, diesel loses its quality over time, so even if we don't use all the fuel, we need to make sure that it is replaced. This item is therefore no longer found in our home bills, but it is absolutely crucial for the data center.

3) Cooling

Another absolutely essential item that you won't find on your home bill, is cooling. All IT technologies generate heat. And that needs to be dissipated. Some data centers cool with liquid, which is highly efficient, however unusable for the vast majority of IT infrastructure. Without modifications, you can lose the warranty or support. Therefore air conditioners are used for this purpose. In our data center we use a cascade of smaller Toshiba Super Digital Inverter air conditioning units with 12-16kW. These units are in redundancy up to N+2 and thanks to this configuration, we can guarantee the quality of the environment. In addition, in 2022 we invested in a free-cooling project, which we are now gradually implementing and will significantly improve the cooling efficiency of the entire data center.

4) Maintenance and amortization

And maintenance, revision and amortization of equipment related to the above items should not be forgotten. So we are talking about revisions, inspections, servicing and depreciation of all diesel generators, UPS, air conditioners, switchboards and so on...

Billing of energy with PUE

Pue meter, effectiveness of power within data center

In the data center environment, we can meet with the abbreviation PUE (Power Usage Effectiveness), which is an indicator of energy efficiency. It is actually a ratio of total power consumption to the consumption of IT equipment and shows how efficient the operation of the data center is. The ideal value is of course 1.0. However, we don't live in an ideal world, and so the really big modern private data centers (e.g. Google) boast an average efficiency of 1.2. It's not unusual, indeed it's correct, that the PUE value is changing during the year, especially with regard to the weather - temperatures.

However, common commercial data centers, where equipment from different customers and vendors are combined, operate with a PUE of 1.3-1.7. In this case, the energy price for the end user in the data center is determined by the amount of raw electricity purchased multiplied by the PUE value.

And here is another bad habit. Many data centers present a very attractive price per kWh in their price list and ads. Unfortunately, many lines below it says that the actual consumption will be multiplied by the PUE value valid for the current billing period. The attractive price per kWh can then become a nightmare.

PUE and Coolhousing

If we use the metric of the non-profit organization The Green Grid, i.e. the mentioned PUE, we work in Coolhousing with an average PUE value of 1.3. In the cold and rainy months of the year we can boast an efficiency of up to around 1.15, while in the dry and hot summer we can reach values up to 1.55.

However, the key point is that the data center client has no chance how to influence or verify the PUE value. He/she can use more efficient and modern equipment, but if the data center does not operate the infrastructure efficiently, it still will not save money. And this aware client will pay the higher price, because the data center passes on its access/non-access as a direct item to the client, who can only influence the PUE by changing the data center.

That is why we in Coolhousing do not use the PUE value in our bills. It is always the actual and, above all, the final price per kWh for the client, which is no longer multiplied by any PUE.