How are artificial intelligence and machine learning poised to change the game for data center operations?

Data centers today face a challenge that seems nearly unsolvable. While data center operations have never been busier, data center operations teams are under pressure to reduce energy consumption as part of corporate carbon reduction goals. Moreover, sharply rising electricity prices are putting budget pressure on data center operators.

With data centers focused on supporting the essential technology services that people increasingly need in their work and lives, it’s no wonder data center operations are so taxing. With no signs of slowing down, we are seeing significant increases in data usage related to video, storage, computing needs, smart IoT integration, and the rollout of 5G connectivity. However, despite increasing workloads, many of today's data center facilities unfortunately do not operate efficiently enough.

This is not surprising given that the average operating life of a data center is over 20 years. Efficiency always depends on the original design of the data center facility and is based on expected IT loads that have long since been exceeded. At the same time, change is a constant, with platforms, device designs, topologies, power densities and cooling requirements all evolving as new applications continue to evolve. The result is that data centers around the world often find it difficult to match current and planned IT loads with their critical infrastructure. This situation will only intensify as data center demand increases. According to analyst forecasts, data center workloads will continue to grow at an annual rate of about 20% between now and 2025.

Traditional data center technologies and methods are difficult to meet these evolving needs. Prioritizing availability largely comes at the expense of efficiency, with too much work still relying on the experience of operations staff and trusting assumptions to be correct. Unfortunately, evidence suggests this model no longer holds true. Research from remote sensor monitoring provider EkkoSense shows that an average of 15% of IT racks in data centers are operating outside of ASHRAE’s temperature and humidity guidelines, and data center cooling energy consumption is as high as 60% due to inefficiency. . This is a major problem, with the Uptime Institute estimating that around $18 billion in wasted energy is lost in data centers worldwide due to inefficient cooling and airflow management. This is equivalent to wasting approximately 150 billion kilowatt hours of electricity.

With 35% of the energy used by data center infrastructure going toward cooling, it’s clear that traditional performance optimization methods are missing huge opportunities to achieve efficiency gains. A survey by EkkoSense shows that one-third of unplanned data center outages are caused by overheating issues. There is a need to find different ways to manage this problem, which can provide data center operations teams with great ways to ensure availability and improve efficiency.

Limitations of Traditional Monitoring Technology

Unfortunately, only about 5% of operations teams currently monitor and report the temperature of their data center equipment on a per-rack basis. Additionally, DCIM and traditional monitoring solutions can provide trend data and be set up to provide alerts in the event of a failure, but these measures are not enough. They lack the analytical skills to gain insight into the causes of problems and how to solve and avoid them in the future.

Operations teams recognize that this traditional monitoring technology has its limitations, but they also know that they simply don’t have the resources and time to take the data they have and derive meaningful insights from analyzing it. The good news is that technology solutions are now available to help data centers solve this problem.

Now is the time to integrate data centers with machine learning and artificial intelligence

The application of machine learning and artificial intelligence has created a new paradigm in how data center operations are handled. Instead of being inundated with too much performance data, operations teams can now leverage machine learning to collect more granular data – meaning they can start to gain real-time access to how their data centers are running. The key is to make it accessible, and using smart 3D visualizations is a great way to make it easier for data center teams to interpret performance and data at a deeper level: for example showing changes and highlighting anomalies.

The next phase is to apply machine learning and artificial intelligence analytics to provide actionable insights. By augmenting measurement data sets with machine learning algorithms, data center teams can immediately benefit from easy-to-understand insights to help support their real-time optimization decisions. The combination of real-time granular data collection and AI/machine learning analytics every five minutes allows operations staff to not only see what is happening in their data center facilities, but also figure out why and what should be done about it.

AI and machine learning-powered analytics can also reveal the insights needed to recommend actionable changes in key areas such as optimal set points, floor grid layouts, cooling facility operations, and fan speed adjustments, etc. . Thermal analysis will also show the best location to install the rack. And, because AI enables real-time visibility, data center teams can quickly get immediate performance feedback on any changes that have been implemented.

Artificial Intelligence and Machine Learning Help Data Center Operations

Given the pressure to reduce carbon emissions and minimize the impact of rising electricity prices, data center teams need new tools if they are to achieve their reliability and efficiency goals. optimization support.

Leveraging the latest machine learning and artificial intelligence-driven data center optimization methods can certainly make an impact by reducing cooling energy and usage – with immediate results within weeks. By putting granular data at the forefront of optimization plans, data center teams are able to not only eliminate the risk of overheating and power failures, but also ensure an average reduction in cooling energy costs and carbon emissions of 30%. It’s hard to ignore the impact this kind of cost savings can have, especially during a time when electricity prices are rising rapidly. Gone are the days of weighing risk and availability for optimization, and artificial intelligence and machine learning technologies will be at the forefront of data center operations.

The above is the detailed content of How are artificial intelligence and machine learning poised to change the game for data center operations?. For more information, please follow other related articles on the PHP Chinese website!