Data Center Trade Survey Highlights Value, AI, and Sustainability Challenges

The Uptime Institute has launched its 14th annual International Data Center Survey, providing a wide-ranging have a look at an trade poised for important transformation.

Performed within the first half of 2024, the survey supplies insights into the practices and challenges confronted by information middle house owners and operators worldwide.

Key findings of the report embrace:

  • Belief in AI for operational choices has fallen to 58%, down from 76% in 2022

  • 33% of operators are growing new capability to deal with high-density cupboards

  • 80% of operators imagine current downtime incidents had been preventable

Operational Prices Emerge as High Concern

In a big shift from earlier years, the Uptime Institute report reveals that value has emerged as the first concern for digital infrastructure administration.

“Actually, should you go to a few of the trade occasions and also you see the quantity of funding going into the sector and the amount of cash that’s been spent, you wouldn’t suppose that is an trade that worries about prices,” Andy Lawrence, government director of analysis at Uptime Institute, defined throughout a webinar detailing the research’s findings on July 31.

Associated:Data Center Catastrophe Restoration: Important Measures for Enterprise Continuity

“However really, the price of constructing and working information facilities after a reasonably sustained interval the place they had been drifting downloaders has gone up within the final a number of years.”

Lawrence attributed this rise to international inflation, provide chain points following COVID-19, excessive vitality costs, employees shortages, and general sturdy demand.

Energy Utilization Effectiveness Stays Steady

The survey revealed that the trade’s common Energy Utilization Effectiveness (PUE) – a key sustainability metric – has remained comparatively steady lately, hovering round 1.55.

PUE has lengthy been a foundational metric for information facilities. In an interview with Data Center Information earlier this 12 months, Peter de Bock, program director of the US Division of Vitality’s Superior Analysis Initiatives Company (ARPA-E), defined why PUE issues as a measure of sustainability.

As to why PUE is stagnant, there are just a few causes.

“Effectivity isn’t the one driving consider an finish information middle design and operations, the opposite factor that’s competing for consideration is availability,” Chris Brown, chief technical officer at Uptime Institute commented. “Availability is job one within the information middle.”

Brown emphasised that information middle operators are continuously balancing the necessity for growing effectivity with sustaining required availability ranges.

Associated:ARPA-E’s Peter de Bock Talks Data Center Cooling Obstacles, Improvements

“As an alternative of fixing for the utmost effectivity that we will get out of the info middle, we now have to unravel for a coincident most between availability and effectivity,” Brown stated. “So, I feel with that actuality we’re not going to see PUE meaningfully lower considerably over the following few years.”

Sustainability general wasn’t precisely a major concern both. The survey highlighted a big hole in sustainability reporting, with lower than half of respondents monitoring metrics like renewable vitality consumption and water utilization.

The necessity to enhance sustainability at information facilities stays a core matter. In July, a coalition of hyperscalers and distributors working collectively within the iMasons Local weather Accord introduced an effort to push for higher transparency in sustainability reporting.

AI Drives Some Excessive-Density Deployments, However Enterprise Functions Stay the Major Driver

The survey indicated a gradual enhance in server rack densities, with extra amenities reporting common densities within the 7-10 kW vary.

AI isn’t but a big driver of elevated information middle density, with solely 23% of respondents citing AI workloads as driving their highest density deployments. The highest driver continues to be commonplace enterprise purposes.

Associated:High Data Center Outage Developments and Methods for Decreasing Danger

Nevertheless, there’s concern that AI might result in extraordinarily high-density rack deployments, doubtlessly exceeding 100 kW per rack. This is able to require custom-built information facilities and important investments in energy infrastructure which may be prohibitively costly for a lot of operators.

Brown commented that information middle operators are struggling to foretell the precise influence of AI on density and capability planning, as there’s a lack of real-world examples to base their forecasts on. This uncertainty is a significant concern for the trade in his view.

Outages Lower, However Monetary Affect Stays Vital

Knowledge middle outages have barely decreased in frequency and severity, however their monetary influence stays substantial. The survey discovered that about 20% of impactful outages value over $1 million.

Energy-related points proceed to be the main explanation for outages, accounting for over half of all important incidents.

“Every little thing in an information middle makes use of energy and energy is pretty binary, it’s both on or it’s off,” Brown stated. “Energy has so many extra vectors by which to trigger your information middle issues and thus energy is the most important trigger and doubtless would be the largest trigger for an extended time frame.”