Intelligent Data Centres Issue 75 | Page 33

E D I T O R ' S Q U E S T I O N
Not being able to understand the needs of the business or the application and translating those needs into practical IT workloads and then again translating those IT performance benchmarks into kW of cooling and power is the most fundamental element of sustainability. So, the gap that depicts the misalignment between what you truly need verses what you own and operate is the core area of inefficiency that needs serious attention.
The incentive – We need to ask ourselves, are people truly incentivised to become efficient? It is an important element and an enabler of suitability to entice organisations to move in the direction of sustainability. Whether you believe in climate change and are a pro-environment organisation or not, sustainability boils down to cutting costs and becoming more efficient and conducive in your operation. If this is properly understood by organisations then sustainability becomes an integral part of their agenda, without oversight. Note that, cutting costs for your organisation could mean serious CapEx and OpEx reductions and for the environment, it all translates to carbon and reversing climate change.
The trend – The most obvious sustainability trend today is buying up more efficient tools and systems. Cooling is a much more complex topic than power in data centres and a bigger challenge to overcome. This is because powering up a rack requires just power, but cooling a cabinet is not as simple as cold air and refrigerant. Inevitably, new cooling methodologies are emerging left and right. The prominent vendors aren’ t necessarily keeping up with all the changes simply because they have less maneuverability and agility than the smaller firms. So one must keep an open mind for the startups in this space.
Intelligence and data gathering is another key topic when it comes to suitability.‘ The more you can measure the more you can manage’, therefore software, sensors and tools that help you sniff more data and provide you more meaningful reports to your dashboard are key trends.
Most critically, the demand for conducting third party assessments to find the true efficiency leaks and sustainability weaknesses is on the rise.
Education is another big piece of the puzzle here. Selecting the latest brochure from your typical vendor doesn’ t make you efficient, or sustainable. Just buying the latest tools doesn’ t cut it anymore and if left standalone, can be counterproductive in the age of AI.
Keep in mind that the chip dynamics and requirements change on a daily basis. This is while the data centre power and cooling technologies cannot keep up with the change. We have inference chips that claim they can be air-cooled at 10kW to 20kW, whereas certain AI chips require 150kW of heat dissipation via being only liquid-cooled and now projecting numbers such as 300kW and even 600kW.
If you just completed the design and development of a 20kW per cabinet aircooled data centre and now the new heat loads become your target audience, you have just wasted your entire investment. This is why certain hyperscalers are slowing down, despite their urge to build further capacities, simply to take a step back to have a more aerial view of where things are ending up as far as chips and kWs are concerned.
So, it all boils down to comprehension and the translation of need and knowing your audience. Thus devising the right data centre around the right set of requirements is truly the key to sustainability in the modern data centre era.
Mehdi Paryavi is the Chairman and CEO of the International Data Center Authority( IDCA), as leading Digital Economy think tank and prime consortium of policymakers, investors and developers in AI, data centres and cloud.
www. intelligentdatacentres. com 33