AI data center optimization benefits – what are you waiting for?

AI data center optimization benefits – what are you waiting for?

Free advisory ebook by Dr Stu Redshaw “Accelerating your time to AI compute benefits” here

When OpenAI launched its text based ChatGPT generative AI system in November 2022 it quickly attracted huge attention, passing its first million users in just five days.

What’s clear is that the ChatGPT launch, major generative AI initiatives from vendors such as Microsoft, Google, Meta, Amazon and Baidu, and hundreds of other generative AI app launches, has put machine learning and AI in the spotlight. Analysts are competing to provide the boldest predictions, Boston Consulting Group, for example, suggesting that generative AI is set to achieve a 30% share of the overall AI market by 2025.

To understand where the latest generative AI tools fit in, it’s first worth establishing exactly what we mean by terms such as AI and machine learning – and how these technologies can be used to support best practice data center optimization.

Machine learning and artificial intelligence are related but distinct concepts in the field of computer science. Artificial intelligence refers to the development of intelligent machines that can perform tasks that would typically require human intelligence, such as visual perception, speech recognition, decision-making, and natural language processing. Machine learning, on the other hand, is a subfield of artificial intelligence that focuses on the development of algorithms and statistical models that enable machines to improve their performance on a specific task over time by learning from data. In other words, machine learning is a tool used to achieve artificial intelligence by allowing machines to learn from experience and make predictions or decisions based on that learning.

Gathering machine learning data at a more granular level
When gathering the machine learning information necessary for data center optimization to be effective, it’s important to collect data at a much more granular level. The concept of garbage in/garbage out clearly applies here, with incomplete, sampled, or out-of-date information inevitably leading to poorer quality optimization decisions.

Our research suggests that just 1 in 20 M&E teams currently monitor and report on an individual rack-by-rack basis, and even less collect real-time cooling duty information. But how granular do you need to get for effective machine learning? At EkkoSense we’re seeing average data centers now being instrumented to produce over 100,000 data points a day, with sensors measuring what’s happening across cooling units, rack temperatures, what the UPS is doing, and exactly what power is being produced – and then bringing all that together in real-time to enable the development of meaningful optimization insights.

The combination of comprehensive monitoring and machine learning algorithms can uncover and continually track the cooling zones within a data center. Introducing powerful algorithms that correlate the relationship between the critical infrastructure and IT load, not only shows what’s happening but also why – effectively enabling data center operations teams to make informed decisions on how to resolve issues.

Optimizing data center performance with AI
It’s only when you combine this level of granular machine learning data with AI algorithms that you can start to translate potentially billions of data points into the insights that can support real-time optimization decisions.

However, if AI is to truly support more effective data center optimization, then it’s essential that the process isn’t complicated. Unfortunately, without real-time visibility of data center cooling, power and capacity information across critical facilities data center teams will always find it hard to identify thermal and power optimization opportunities – let alone act on them. That’s where AI-enabled intuitive 3D visualizations can make a huge difference. The application of the latest digital twin models coupled with remote access to heartbeat M&E data helps to make the expertise behind machine learning algorithms and AI analytics as intuitive as possible.

AI-powered data center optimization allows data center teams to make informed decisions on how to resolve potential issues. The latest AI-enabled software can observe changes in real-time and will often inform you that a failure is going to occur before an issue has even been flagged by traditional BMS systems. AI can change the game for data centers, allowing teams to visualize airflow management improvements, manage complex capacity decisions, and quickly highlight any worrying trends in cooling performance.

At EkkoSense we advocate a light-touch approach to AI-powered DCIM, equipping teams with the human auditability that’s so important to them. While powerful correlation engines learn how your data center is operating and show what can be done to improve things, we always leave ultimate determination and control with the human operator. Actionable recommendations could include optimum cooling unit set point adjustments, fan speed points and standby settings, changes to floor grille layouts, as well as advice on optimum rack locations. This approach to AI ensures that control and accountability stay with the operator – but with all the benefits of the guesswork being removed from the decision process. This also recognizes that operators will often have insights that simply aren’t available to the AI data sets.

And the benefits of this approach are clear, with reduction in thermal risk, cooling energy savings, leaner running through more effective capacity planning, and quantifiable carbon savings all achievable through the application of machine learning and AI techniques. At the same time, the ability to learn from the past supports detailed trend analysis and a much more proactive risk avoidance strategy.

Unlocking the promise of AI in the data center
The release of applications such as ChatGPT and other generative AI tools that employ deep learning algorithms to identify patterns and features within data sets are expected to have a profound impact on the global software industry.

The one thing that’s certain is that the results currently generated by ChatGPT are only going to get better given that AI capabilities are improving at an exponential rate. For example, the paragraph italicized at the start of this blog defining machine learning and artificial intelligence was entirely the work of ChatGPT, and the output from these kinds of tools is becoming hard to distinguish from original content.

Unfortunately, good enough doesn’t cut it in the data center space. Using generative AI to create paragraphs of text is an interesting experiment, but organizations simply aren’t ready to hand over control of their critical facilities to AI. And at a time when the industry is already under pressure to reduce power consumption, there’s also a very real question around whether organizations can actually afford the huge compute and storage needed to support generative AI’s increasingly large models.

That’s why at EkkoSense we’re focused on providing a distinctive, light-touch AI-enabled software-driven thermal optimization approach that enables operations teams to optimize their data center performance while simultaneously delivering quantifiable sustainability results.

The result is data optimization that has been developed and refined to be particularly easy for data center teams to implement, understand and use in-house. Unrivalled levels of IoT sensing bring an entirely new class of accuracy and granularity to data centre operations – providing the core machine learning data that enables true real-time visibility of cooling, power and capacity performance. Our EkkoSoft Critical AI-powered optimisation software then provides the 3D visualization and analytics suing powerful AI algorithms that correlate the relationship between critical infrastructure and IT load. Observing changes in real-time means you can work to remove thermal and power risk, optimize data center cooling capacity, minimize energy waste, and contribute directly to your corporate ESG goals. Contact me to continue the conversation [email protected] or download our free advisory ebook by Dr Stu Redshaw “Accelerating your time to AI compute benefits”