KUALA LUMPUR, Dec 11 (Bernama) -- The ever-expending digital economy is driving the growth of data centers in Malaysia. Digital transformation of businesses into online businesses such as in retail, banking, insurance, and more are one of the causes. In addition, the growth of artificial intelligence (AI), video streaming, storage, and backups, 5G, internet of things, digital payment, and more are pushing data centers development into a gold rush of the 21st century. The growth of data centers in Malaysia (and surrounding region) is further driven by the spillover effect from Singapore. Singapore limited data centers developments after achieving a milestone of 7% of the nation energy being supplied to data centers alone.
It is common for big corporates to build and operate their own data centers. However, most other corporations chose to locate their servers with a co-location data center. We can imagine a co-locator as a long-term hotel for servers. Corporations rent spaces from these co-location providers. Organizations have shifted to co-locators because they can offer better security, redundancy, and services, while allowing them to focus back on their core businesses.
Meanwhile, the recent growth in awareness of Net Zero and ESG is pressuring all data centers to go green. Clients of co-locators are organizations that have made commitments to Net Zero and ESG. To keep these customers, data centers must meet their needs. Unfortunately, data centers are high carbon emitters and high in water consumption. It is a highly polluting industry.
While the demand for data centers continues to soar with the expanding industry landscape, we cannot turn a blind eye to the fact that traditional data center models are inherently energy-intensive. It is incumbent upon us to champion a paradigm shift towards sustainable data infrastructure.”
According to the Green Building Index (GBI), the immediate goal in data centers today is improving energy efficiency. The most important metrics for measuring the energy efficiency of a data center is PUE, or power usage effectiveness. PUE is a ratio of the total amount of power used by a data center to the power used by the IT equipment. A PUE of 1.0 is ideal, meaning that all the power used by the data center is used to power the IT equipment. A PUE of 2.0 means that half of the power used by the data center is wasted on non-IT equipment.
PUE=(Total Facility Power)/(IT Equipment Power)
The goal is to reduce power consumption of non-IT equipment as much as possible, to bring PUE close to 1.0. The latest data from Google¹, showed that a few of their data centers are operating with PUE as low as 1.07 in 2023. Worldwide, Google data centers have a global average of 1.10.
The Uptime Institute is a recognized certifying body for data centers, and they regularly publish data on data centers’ PUE. In 2007, they published that the global average PUE in large-scale data centers was 2.5. Today the average PUE has reduced to 1.46 in Europe, 1.55 in North America, 1.60 in Asia Pacific, and 1.79 in Middle East & Africa.
The Singapore government has recently set a PUE limit of 1.3 for new data centers. Although rare, there are already data centers in Singapore with operating PUE below 1.3 today, with one published to be operating at PUE of 1.1 using liquid cooling technology.
In a benchmarking report published by Singapore in 2012, the average PUE in Singapore was 2.07. In the same report, they have also summarized that approximately 50% of power consumption in a data center is due to the IT equipment, 37% due to the cooling equipment, and 12% is for rest of electrical consumption such as lighting, power distribution, UPS, ventilation, and the dedicated office spaces. In short, the biggest chunk of non-IT energy consumption is used for cooling of IT equipment. The energy consumption of office spaces is relatively small in an overall picture of a data center.
Although Malaysia have not set any target on energy efficiency for data centers, ESG and Net Zero movement will drive energy efficiency in this industry. For example, YTL announced last year that their 500 MW data center campus in Johor will be powered by solar energy, and it will have a PUE of 1.2.
Air-Cooling in Data Centers
Air-cooling is the conventional way of cooling data centers. In the past, it was common for data centers to be operated at extremely cold temperatures. Often temperatures were set below 20⁰C in the assumption that it was good for the IT equipment. After extensive research, ASHRAE published a document in 2008, revealing that IT equipment can operate in the range of 18⁰ to 27⁰C without any issues.
This revelation droves innovations in air cooling technology for data centers. Designs of air-cooling system were revamped for high-temperature cooling. Data centers were built with hot-aisle cold-aisle configuration, displacement ventilation, in-rack cooling, and more. These innovations drove the energy efficiency improvements in data centers till today. However, we may be approaching the tail-end of air-cooling for data centers.
Liquid-Cooling
Due to the increasing power densities of servers, where above 100 kW per rack is expected (typically it is around 6 to 15 kW per rack today), there is a need for a major revamp of the cooling system for data centers. Conventional air-cooling will no longer be enough to provide the cooling needed by such a load. Liquid-cooling is required.
More interestingly, liquid-cooling has the potential to be significantly more energy efficient than air-cooling because water has 4x higher thermal capacity than air. It will take 4x less energy to move heat out of the building compared to air-cooling. Designed well, the domino effect from such an advantage can led to drastic reduction in energy consumption to a data center.
Currently, technologies on liquid-cooling have split into two directions, cold-plate, and immersion.
Cold-plate is a water-cooled copper plate directly attached on top of a processor chip. Initially, there were concerns of water leakages, but these concerns were taken away by operating the water circuit below the atmospheric pressure (if there is a leak, air sucks in, without leaking water out). There are also many other proprietary technologies in the market today addressing this concern.
Immersion cooling is the use of liquid that does not conduct electricity. Therefore, the entire circuit board can be immersed into this liquid without short-circuiting. 3M used to provide such proprietary liquid but has recently stopped producing it. The 3M technology was interesting because it worked on a heat-pipe principle, where the heat from the servers boils the liquid into gas, transporting it without adding energy for it to be cooled into liquid again. The cooled liquid flows back down to the server for the cycle to repeat again, mimicking our natural rain cycle.
Another liquid that is commonly used for immersion cooling is mineral oil. Mineral oil does not conduct electricity, allowing any circuit board to be immersed in it. But this idea has yet to gain mainstream although it has been around for some time. It is still an unconventional idea for the industry to imagine an entire circuit board (server) can be submerged in a liquid.
Water consumption
In Malaysia, water consumption of data centers is primary driven by the need of cooling. All the heat generated in a data center is typically removed by a cooling tower via evaporation of water. As the power densities of data center is projected to increase, the amount of heat to be rejected will also increase and along with it, the evaporation of water.
The Water Usage Effectiveness (WUE) is commonly used in Singapore. It is a measure of the water efficiency of data centers. It is calculated as the amount of water used by a data center divided by its annual operating IT equipment load. The lower the WUE, the more efficient the data center is in its water use.
The average WUE for data centers in Singapore is 2.6 m3/MWh. This means that, on average, data centers in Singapore use 2.6 cubic meters of water to operate one megawatt hour of IT equipment. The global average WUE for data centers is 3.5 m3/MWh.
There are data centers in Singapore with WUE less than 2 m3/MWh. These data centers are using a variety of water efficiency measures, including:
· A more efficient cooling system.
· Rainwater harvesting and recycling of water.
· Waterless cooling technologies.
Singapore is actively funding projects to develop a waterless cooling technology for data centers. This technology could potentially revolutionize the data center cooling industry and reduce water consumption even further. At home, our split unit air-conditioning system is air-cooled, without a water-based cooling tower. Therefore, there is a possibility of making this happen in data centers.
GBI Rating for Data Center
Two tools in GBI can be used for data center. New constructions will use the NRNC (Non-Residential New Construction) tool, while existing data centers can use the EB (Existing Building) tool. The only difference in GBI between a typical building and that of a data center is the scoring for Advance EE Performance, where PUE is specified instead of BEI (building energy intensity). The credits scoring for PUE in GBI is as follows:
PUE Points
< 1.9 2
< 1.8 3
< 1.7 5
< 1.6 8
< 1.5 10
< 1.4 12
< 1.3 15
The above points scored for each PUE level were established many years ago. Based on the report from the Uptime Institute, the PUE in Asia Pacific averages at 1.60 today. In addition, it is also known that Singapore has set a limit of PUE at 1.30 for new data centers. This may be a suitable time to revise GBI to start scoring credit at PUE of 1.60 and to end at PUE of 1.30. The smaller interval between PUE to earn additional credits will encourage projects to push ever slightly more to gain additional credits.
In addition, GBI may also consider a ‘reducing interval’ for points scoring, where the higher PUE has a bigger interval for additional credits, while at lower PUE, the intervals become smaller for additional credits. This will make the system fairer, because it is easier to improve energy efficiency when it is not efficient and becomes more difficult to make improvements when it is already efficient.
In GBI, water efficiency is addressed by rainwater harvesting, water recycling, efficient irrigation/landscaping, and metering & leakage detection. The water used by cooling need is indirectly addressed by these requirements. However, for a data center, it may be necessary to address water consumption for cooling directly because this is the overwhelming use of water in a data center. The WUE is a simple measurable indicator of water efficiency in data centers that is globally recognized today. It can be a useful indicator for awarding credits on the water sustainability of a data center.
Due to climate change, there is a strong possibility of drought (and flood) in Malaysia depending on El Nino or La Nina weather cycle. It is worthwhile preparing our data centers to be ready for such events by reducing water consumption, thereby reducing the risk of climate change on their business.
GBI addresses sustainability issues beyond energy and water. For example, there is an entire section dedicated to indoor environmental quality on air, view, and acoustic comfort for building occupants. In addition, GBI also addresses various environmental issues on site, before, during, and after construction. While these sustainability issues are important, the primary pollution from a data center is carbon emissions and high-water consumption. Moreover, the office spaces (and number of people working) are small in a data center. Due to this, there might be an opportunity to fine-tune GBI weightages for a data center in Malaysia to better reflect its sustainability need. A GBI tool specifically addressing the need of data centers for our Malaysian environment, will encourage more data centers to adopt it and put GBI ahead of other green building ratings tools for data centers.
Conclusion
It is forecasted that green data centers will be growing exponentially due to digital transformation, Net Zero, and ESG movement. The overwhelming power consumption is the primary driver of pollution for data centers in Malaysia, namely high carbon emissions and high-water consumption.
PUE and WUE are two commonly used indicators of energy efficiency and water efficiency in data centers. However, these indicators do not fully address all the sustainability issues in data center development. Green building ratings such as GBI can help to address these gaps. However, there may be room to improve the weightages in GBI to closely reflect the sustainability impact of data centers in tropics.
“GBI recognizes the urgency to address the environmental impact of data centers. We are committed to ensuring that our rating tools evolve to meet the specific needs of this burgeoning sector,” said GBI Accreditation Panel Chair, Ar. Dr. Serina Hijjas.
The drive for green data centers is an opportunity for GBI to innovate a win-win outcome for the environment, society, and economy in data centers. GBI can be a leading tool by addressing the need of ESG and Net Zero to mitigate climate change and in reducing climate change risk in data centers.
¹Source: https://www.google.com/about/datacenters/efficiency/ and Green Building Index
GBI Backgrounder
GBI is Malaysia’s industry-recognised Green Rating Tool for buildings and townships. It is a wholly-owned subsidiary of Pertubuhan Akitek Malaysia (PAM) and the Association of Consulting Engineers Malaysia (ACEM). For more information, please visit www.greenbuildingindex.org or contact:
Source: Greenbuildingindex Sdn Bhd
FOR MORE INFORMATION, PLEASE CONTACT:
Name: Nur Aqilah Husna
Tel: +6013 4407375
Email: aqilah@greenbuildingindex.org
Name: Andrew Kong
Tel: +6012 6818838
Email: andrew@greenbuildingindex.org
--BERNAMA
No comments:
Post a Comment