The hidden impact of cloud storage

This is blog 4 in the blog series on Green IT.

The previous blog post looked at the immense growth of digital data. We also saw that a vast amount of these data (some 80 per cent) are never again used after being created. In this blog we will examine the cloud, which is where these data usually end up. We will also look at the ecological impact of this omnipresent storage method on the climate.

Wolf in sheep’s clothing

While a ‘cloud’ conjures up an innocent image, data centres in fact consist of endless rows of humming servers, accessed through a network infrastructure. These data centres operate 24/7, every day of the year. Several studies, including the 2016 study Optimizing energy consumption for data centers (1) by the Chinese researcher Huigui Rong and his team, reveal that these centres consume a staggering amount of energy.

The high energy consumption is mainly due to the use of electricity for IT equipment (servers and infrastructure) and cooling equipment, as well as for lighting, power distribution and other demands. The cooling system alone can account for 40 per cent of the data centre’s energy consumption, on average (2).

Was everything better in the past?

Before cloud technology emerged, data were usually stored in-house (or ‘on-premise’), using local servers in special server rooms. Was that more eco-friendly? Not really. The article Cloud infrastructure could be key to meeting climate goals (3) shows how cloud storage is on average up to five times more efficient than storage on so-called ‘on-premise’ servers.

According to a study into cloud storage in the Asia-Pacific region, going from local storage to cloud storage could result in a 78 per cent reduction in energy use and the associated CO2 emission (4). Research by 451 Research even showed that organisations in Europe could reduce their energy consumption by almost 80 per cent by shifting their data processing from local data centres to the cloud (though it should be noted that this was research commissioned by Amazon) (5).

Similar research by Microsoft (which could again be an instance of the fox guarding the chickens) comes to the same conclusions. Together with WSP USA, they examined how the transition from on-premise storage to the cloud can reduce the CO2 footprint. Result: cloud solution Azure is 93 per cent more energy efficient and causes 98 per cent less CO2 emission than on-premise servers. (6)

Using server capacity more efficiently reduces energy consumption

The significant difference in energy consumption and CO2 emission between cloud and on-premise storage is due to the fact that data centres achieve a much higher and therefore more efficient use of servers than local server parks. According to estimates by the American Environmental Protection Agency, on-premise servers usually utilise just 5 to 15 per cent of the server capacity, while servers in the cloud are often utilised for 50 per cent or more. (7)

Highly innovative sector

Furthermore, large cloud providers lead the field in the development of energy-efficient data storage, as they are well aware of the climate impact. Research into innovation focuses on various aspects, such as reusing the water used to cool servers for the heat grids servicing nearby households. Examples are the heat grid built in the new neighbourhood of Smariuskade in the Dutch city of Tilburg (8), or the heating of student dwellings in Amsterdam (9).

Additionally, robust investments are being made in solar and wind parks so that data centres can operate fully on green power, such as the wind park in Wieringermeer (10) for Microsoft’s data centres. Another approach is to see how the storage capacity of existing data centres can be boosted further, also by looking into alternative storage methods (more about this in a next blog).

With this innovation drive, the sector is leading the field in terms of energy transition. The sector expects to be fully CO2-neutral by 2030 in terms of energy use (11). Finally, researchers are examining the use of hydrogen as source of energy and the storage of energy in battery parks as a way to unburden the electricity grid (12). These are issues that still go unanswered in this time of climate transition.

The Jevons paradox

Now you might want to lean back and think that things aren’t so bad after all, but that’s not the way it is. This is where the so-called Jevons paradox crops up. This is an ecological-economic concept that states that improvements in the efficiency of using a resource (such as fuel and energy) do not necessarily result in a reduced use of that resource. In fact, we often see that it leads to even greater use.

In 1865, the Englishman William Stanley Jevons predicted that when the use of coal would become more efficient in the United Kingdom, this would make the use of coal more attractive and cause the demand for coal to increase rather than decrease.

Applied to the IT sector, we see that the ever-advancing technology goes hand in hand with an exponential increase in the amount of stored data. So the problem is actually just getting worse all the time, certainly if individual users do not take any action.

Considering the huge amount of dark data (as much as 81 per cent), we should be able to save massive amounts of energy by being more critical of what we do and do not save long-term. Additionally, the scarce green power that is now powering data centres could then be used to green other sectors or domestic households.

Checklist green/greener data centres

If you are considering storing data in the cloud, or switching to another cloud provider, here are a number of points you could include in your process of selecting a provider:

Energy usage:

o Type of power: does the data centre use green or grey power? If a data centre uses green power, then it should demonstrate this with a green certificate that indicates the origin of the power.

o How high is the power usage effectiveness score (PUE value)? This metric is an indicative value showing how efficiently a data centre uses power (13). The value is calculated based on the amount of energy that is used for peripheral systems such as cooling and emergency power supply (UPS). A PUE value of 1.25 means that, for every 100 Watt of power used by a server, 25 Watt is used for peripheral systems. Worldwide the score averages 1.59, in Europa it is 1.46, and in the Netherlands it is even lower (around 1.3).

o Eco mode and use of green software: this responsibility is generally carried by the data centre customers (so it’s something to ask your Saas (software as a service) supplier about).

▪ Does the centre apply server software optimisation?

▪ Does the centre use energy efficiency functions in existing servers (for instance a sleep mode)?

Water usage:

o How high is the water utilisation effectiveness score (WUE value)? This metric shows how much water a data centre needs to maintain its desired ambient temperature, under conditions of a certain outdoor temperature and humidity.

o Where is the cooling water sourced from? Does the data centre use raw water (not of potable quality) or potable water? Does the data centre pump up and clean its own water from sewage purification plants and water basins?

o What happens with water that is used to cool the data centre? Is it recycled, and if so, are any chemicals used in that process?

Overall environmental impact:

o Is the data centre in possession of an ISO 14001 certificate? Organisations with this certificate have a policy in place regarding environmental management. They optimise their processes and systems to minimise their negative impact on the environment, they obey the laws and regulations, and they treat improvement on these points as a continuous process.

By considering these criteria when selecting a cloud provider, you can limit the CO2 impact of the data you really do need to store long-term. The next blog looks at alternative long-term storage media that are already available now or will become available in the future.

Sources

(1) Rong, Huigui et al., ‘Optimizing energy consumption for data centers’, In: Renewable and Sustainable Energy Reviews, Vol. 58, May 2016, pages 674-691.

(2) Avgerinou, Maria et al., ‘Trends in Data Centre Energy Consumption under the European Code of Coduct for Data Centre Energy Efficiency’ JCR Technical Reports, 2017.

(3) Sambhi, Sonia, ‘Cloud infrastructure could be key to meeting climate goals’ published on 15 October 2021.

(4) iBestuur, ‘Waarom overstappen naar de cloud onderdeel zou moeten zijn van uw duurzaamheidsstrategie’, published on 22 February 2022.

(5) Solimas, ‘Hoe de cloud bijdraagt aan een beter klimaat’ references the report ‘The Carbon benefits of cloud computing’.

(6) NDRC Issue brief, ‘Is Cloud Computing Always Greener?: Finding the Most Energy and Carbon Efficient Information Technology Solutions for Small and Medium Sized Organizations’, published in October 2012.

(7) S&P Global Market Intelligence Black&White, ‘The Carbon Reduction Opportunity of Moving to the Cloud for APAC’ published in July 2021.

(8) Vodafone, ‘2025: VodafoneZiggo datacenter verwarmt woonwijk in Tilburg’, consulted on 5 October 2023.

(9) Nikhef, ‘Datacenter Nikhef verwarmt nog eens 720 studentenwoningen’, published on 14 November 2019.

(10) Tweakers, ‘Microsoft neemt tien jaar lang alle energie af van nieuw windmolenpark Nuon’, published on 2 November 2017.

(11) Dutch Data Center Association, ‘Datacenters & Duurzaamheid 2021 : duurzame groei door de digitale infrastructuur’, published in April 2021.

(12) The study into the storage of green energy is also described in the above-mentioned report by the Dutch Data Center Association.

(13) We should add the critical note here that the PUE value mainly reveals something about the data centre’s owner’s approach to sustainable entrepreneurship. After all, the metric does not indicate the total amount of power used by the data centre. An insightful article on this issue is: ‘PUE: Cold Con-Fusion in the Data Center’.