É«¿Ø´«Ã½

By Patrick Smith, VP, EMEA Field CTO, É«¿Ø´«Ã½

Today, energy costs are truly unpredictable, and the rise of AI is going to add to the uncertainty. Today's data centers are already struggling to source enough electricity, while power-hungry GPUs are set to push per-rack kW needs to unprecedented levels. Meanwhile, data centers guzzle water and struggle to cool colossal sheds of hardware, and e-waste piles up as servers and storage become obsolete and are forklifted out to landfill.

It's a bleak outlook, but there are ways to offset energy consumption, water and cooling requirements, as well as e-waste. There are several key contributors to this challenging environment including servers and their CPUs and GPUs, the network, and the storage hosting rapidly growing data volumes. In this piece we'll look at how to achieve greater efficiency, meet sustainability goals, and reduce operating costs with a modern all flash data storage platform, that avoids obsolescence with an architecture that allows it to remain modern for 10 years or more.

According to the IEA, in the US, data centers account for around six percent of all power use, concerning when you consider that 33% of the world's data centers reside in the country. In Ireland, the situation is even more worrying, where by 2026, data centers could account for 32% of all power consumption due to a high number of new builds planned.

Among the factors pushing increased energy usage in the data center is AI, which depends on massive compute power via GPU servers. In fact, the IEA also estimates that data centers, cryptocurrencies, and AI power usage represented almost two percent of global energy demand in 2022 -- and that these energy demands could double by 2026.

Meanwhile, customers are looking for lower power and cooling costs together with lower GHG emissions in the face of increasing regulatory pressure to meet sustainability targets. In Germany, for example, the Energy Efficiency Act requires organizations to report their goals and be carbon neutral by 2045, with real progress to be shown by 2030. Of course, there is also a moral obligation to reduce energy consumption, which has an impact on the planet, a company's reputation, and ability to recruit the next generation of talent. Employees often don't want to work for companies that put profit above all else.

Data centers need to run at temperatures between 18-27 degrees Celsius. Allowing temperatures to rise beyond this will lead to disastrous results, fast. To maintain these temperature levels, they must consume electricity to cool the equipment. Data centers also consume a lot of water, which is set to become an even bigger problem -- especially when you take into consideration certain arid environments such as California, which are also rife with data centers, leading to competition for water between people and technology. Many data centers utilize evaporative cooling, spraying water onto cloth strips, which evaporates, cooling the air around it. It's a smart idea but it's problematic, given the added strain that climate change is placing on water resources -- especially in built-up areas. As a result, this method of cooling has fallen out of favor in the past year, resulting in a reliance on more traditional, power intensive cooling methods like air conditioning.

According to the IEA, the three major hyperscalers release water consumption figures. In recent years these have included Google's use of 450,000 gallons per day in 2021, and Microsoft's total usage of 1.7 billion gallons in 2022, a 34 percent increase from 2021. The average data center uses 1.8 liters per kWh. This is not only bad for the environment generally, but it can be detrimental to areas of the world where water is in short supply, especially urban centers.

E-waste is a huge issue globally. According to the World Health Organization, e-waste is the fastest growing solid waste stream in the world, increasing 3 times faster than the world's population. Not only this, but E-waste contains multiple known and suspected neurotoxicants, including lead and mercury, that may disrupt the development of the central nervous system during pregnancy, infancy, childhood and adolescence.

About 40 million tons of electronic product waste is generated every year, with most heading to landfill, only about 12.5% recycled and 70% of it being toxic. In the data center, the main source of e-waste derives from the succession of 3-5 year hardware refresh cycles that are traditional in IT. In such scenarios, typically, tons of server and storage hardware are forklifted out to be replaced by similar plastic-adorned boxes with different and newer internal components.

Data volumes continue to grow driving deployments of more storage hardware. In this area one clear solution to these challenges is the adoption of flash technology across the data center, specifically high-density flash storage modules. Flash-optimized storage systems use between 2-5x less power than commodity SSD-based systems and between 5-10x less power than HDD-based systems. Some vendors can reduce energy usage and carbon emissions by up to 85% compared to competing flash offerings. along with 3x industry average SSD reliability and 6x the reliability of HDDs.

More efficient flash storage doesn't just reduce power consumption, it also directly leads to lower data center cooling and water requirements. Some companies have engineered their arrays to get the absolute most out of flash. For example, some vendors leverage "raw" flash to build their flash arrays, rather than relying on buying commodity solid-state drives (SSDs), which talk to their flash drives in essentially the same way they would a legacy hard drive. This maximizes the capabilities of flash and provides better performance, power utilization, and efficiency resulting in up to 95% less rack space than competitor products, especially those based on spinning disk. That all means less power, cooling and water requirements in the data center.

It also pays to look at ways to reduce waste in the technology lifecycle. Upgrading legacy storage architecture involves a full forklift upgrade to completely replace old hardware. Handling this process end-to-end is a complex undertaking, requiring IT teams to manage physical waste disposal, overcome security concerns and ensure that downtime is minimized. This model is fundamentally incompatible with achieving sustainability goals and reducing e-waste. Non-disruptive upgrades are one alternative designed to avoid obsolescence, with component modules that can be swapped out non-disruptively for 10 or more years. That compares with the typical storage hardware lifecycle of 3-5 years before reliability and support become significant risks. It's also worth considering how reliability contributes to lower e-waste, flash optimized storage modules typically have 3x the industry average SSD reliability and 6x the reliability of HDDs.

Lastly, if you can't see a problem, it can be difficult to address it. As such, companies should look for storage vendors that can provide transparent monitoring of their storage deployments which gives visibility into energy consumption and provides organizations with recommendations on how to maximize energy efficiency.

So when considering new storage to meet the demands of ever growing data volumes it's worth factoring in the impact on the data center in terms of power, cooling and space but also the environment and targets to reduce GHG emissions and e-waste.