Rawpixel.com - stock.adobe.com

Will AWS pledge to extend life of servers inspire other cloud firms to follow suit?

Amazon’s recent announcement to extend the useful life of its datacentre equipment has been deemed a positive decision for the cloud computing giant, but will other companies adopt this cost-saving and sustainability strategy?

Amazon Web Services (AWS) made a surprise announcement during its February 2022 financial results about its plans to extend the useful life of the servers powering its cloud from four years to five, and its networking equipment from five years to six.

During an analyst call, Amazon chief financial officer Brian Olsavsky said the change in strategy would allow the company to ensure the servers and networking equipment in its datacentres last longer.

“We’ve been operating at scale for over 15 years and we continue to refine our software to run more efficiently on the hardware,” he said. “This then lowers stress on the hardware and extends the useful life, both for the assets that we use to support AWS’s external customers as well as those used to support our own internal Amazon businesses.”

Overall, the move will enable AWS to decrease asset depreciation expenses and improve sustainability across its business, he said.

But what does the wider industry think of this announcement, and do they think other cloud providers will be encouraged by the moves Amazon is making to unveil similar plans?

The general industry consensus is that AWS has taken a positive step. Max Schulze, executive director at the Sustainable Digital Infrastructure Alliance (SDIA), calls it a “good move” that will help AWS save money and reduce resource consumption as it looks to become a leaner and greener datacentre operator.

“Servers last a lot longer than the financial accounting suggests, [around] three-to-five years, and since performance is not increasing as fast anymore – [due to the limitations of Moore’s Law] – why bother with constantly upgrading equipment?” he said. “The longer it can be used, the more profitable it becomes for AWS and the less resources are consumed to manufacture new servers.”

Economic decision

In terms of why AWS made this call, Schulze is of the view that it’s a “purely economic decision” but stresses it will positively impact the company’s sustainability and efficiency goals in the long-term. “As a cloud provider, I utilise hundreds of thousands of servers, I can see if they are degrading over time and assess if there is an increased outage risk,” he said.

“It appears that AWS has figured out there is neither a degradation nor an increased risk. AWS has always been a leader and first-mover when it comes to optimising costs and unit economics – that’s the core of their business, so this is not surprising.”

Amazon’s decision to increase the useful life of its servers and networking equipment can be viewed and assessed from several different sustainability-related viewpoints.

There are circular economy and energy consumption benefits to be gained from this strategy, and it may also contribute to a reduction in the greenhouse gas emissions (GHG) generated by Amazon, said Jay Dietrich, research director of sustainability at the Uptime Institute.

In terms of circular economy benefits, extending the lifespan of datacentre equipment will help the company decrease its refresh cycles from three to two over 12 years. “This saves one cycle of equipment buys and reduces the material use and manufacturing, transportation, and disposal GHG emissions associated with a server buy,” said Dietrich. “Server costs are estimated to range from $6,000 to $10,000.”

This decision will also allow AWS to lower its overall energy consumption and operating GHG emissions as it attempts to go net-zero carbon by 2040.

“A new server can deliver more work per each unit of energy, requiring fewer servers to do the same amount of work,” he said. “Data has shown that each generation of servers can do 30% to 150% more work per server. For each server refresh, assuming the same average server (or network equipment) utilisation rate is maintained, less equipment and energy will be needed to deliver a given workload. A shorter refresh cycle will reduce the energy use required to satisfy a given workload.”

Reputational benefits

Amazon has set ambitious goals to tackle carbon emissions, and Chris Weston, principal of client advisory at IDC, is confident the company is heading in the right direction by seeking to extend the useful life of its servers and networking equipment. “Sustainability is always on the agenda of CIOs [chief information officers] and organisations are keen to manage their supply chains as effectively as possible to show they are not simply passing off their problems to others,” he said.

“We see this through the interest in the IDC Sustainability Index, which measures technology vendors over a range of factors. This move will certainly reduce barriers to buying where people are taking this wider view on supplier selection, and will do Amazon no harm at all in that area.”

What is harder to predict, said Dietrich, is whether this announcement will bolster Amazon’s reputation from an environmental perspective, as operating a sustainable datacentre involves making a lot different moving parts fit together. “Sustainability encompasses the elements of datacentre siting, design and construction, facility and IT management and energy efficiency, renewable energy procurement, GHG emission reductions, water management and use efficiency, [as well as] circular economy considerations – the extension of product life, recycle and reuse of end of life equipment, and reuse of datacentre heat,” he told Computer Weekly.

“There is no single accepted or recognised standard that addresses all these elements – different groups weigh the importance of the various elements differently. This makes it nearly impossible to assess how one single action on one single element will affect an operator’s reputation.”

Dietrich claimed there is a disparity between the sustainability and cost benefits of longer refresh cycles and the sustainability impact of lower energy use and GHG emissions. But generally, he said the industry will react positively to Amazon’s desire to increase server and networking useful life.

If the industry continues to prioritise zero-emissions energy, he added, the advantages of emissions reductions from reduced energy usage may potentially deplete. “This could suggest the sustainability and cost benefits rest with extending the refresh rate,” said Dietrich.

“However, most datacentre operations claiming 100% renewables or moving to a near-term 100% renewable goal are consuming only 10% to 60% renewables in their operations – the rest of the zero-emissions claims are underpinned by unbundled Renewable Energy Credits (RECs) matched to grid purchases,” he said. “These datacentres still have energy-based GHG emissions connected to their energy consumption and the reduced operating GHG emissions are a relevant consideration.”

Industry change

With Amazon choosing to increase the lifespan of its datacentre infrastructure to lower operational costs, boost sustainability and become more efficient, will other enterprises and large-scale cloud firms follow suit?

In response to this question, Yugal Joshi, partner at Everest Group, points out that Amazon is not the first major cloud provider to make this type of decision.

“The initiative to extend the useful life of servers was done by other cloud vendors, such as Google, too,” he said. “Of course, it directly saves cost and does positively impact the environment.”

But he also warned it may be difficult for other enterprises and datacentre suppliers to implement similar strategies, as cloud giants such as AWS have created technology processes that cannot be easily replicated.

“Moreover, enterprises will have to change their depreciation policies, budgeting cycle and multiple other aspects to make this change happen,” he said. “These changes go beyond just technology and impact people, processes, and risk and supplier management. Therefore, the overhead of doing something like this with an uncertain success rate can be a big challenge for enterprises and other DC vendors.”

Read more about datacentre sustainability

Weston echoed these sentiments, calling useful life increases a “balancing act for everyone involved”. In particular, he warned that maintaining datacentre equipment for longer may result in more work for technologists.

“With care around maintenance and resilience, outages should not be the issue – it’s more a question of coping with the predicted increased workloads,” said Weston. “For most enterprises, data will be growing at 40%-50% over the next five years, both on-premise and in the cloud.

“IT budgets will grow at around 10%-12% a year. We are seeing many organisations look to move to denser workload consolidation to avoid unwanted increases in energy and floorspace for tech, and improve server and storage utilisation. However, to achieve this, an investment in newer, more software-defined storage systems is necessary.”

Dietrich, on the other hand, said extending refresh times is an operator-dependent decision that varies by workload type. “The decision will depend on expected failure rates, whether a single server failure can be absorbed in the server infrastructure (public cloud) versus an enterprise server with only a single or small group of servers where a failure could have high business impact, the importance of increased capacity and energy efficiency to a given workload such as high performance compute, and other considerations.”

Jason Gregson, global head of AWS programs and operations at DoiT International, said extending the useful life of servers and networking equipment will only work if it doesn’t affect their usability. “The move will certainly lower Amazon’s capital expenditure and therefore improve its profit margins – however, it’s true to say there are also potential advantages from a sustainability standpoint,” he said.

“For AWS customers, it’s unlikely that the cost reductions associated with using older equipment would be passed on, but as long as SLAs are adhered to and they don’t start to see more frequent issues and outages, that is unlikely to matter much,” said Gregson. “If, however, we do see an uptick in outages, then customer trust will erode quickly and attrition to alternative providers could become a real risk. The success or failure will be shown in the data, and how Amazon reacts if things don’t quite go to plan.”

With sustainability and cost savings high up on the corporate agenda, it’s no surprise that Amazon has decided to increase the useful life of its datacentre servers and networking equipment. Although this move certainly shows that the company is serious about its responsibility to decrease carbon emissions and improve efficiencies over time, there are challenges that might make it less appealing to other cloud operators.

Read more on Datacentre energy efficiency and green IT

Data Center
Data Management