Managing and enabling the operation of data centers is safer than it was not too long ago, but it is still far from being considered a safe place for workers who have to manipulate components or update and install servers. Basically, because they have to work in an environment where they are exposed to electrical breakdowns or leaks of chemical components, among other safety problems. That’s why the leading cloud providers, who are also among the top data center owners and operators, are looking for solutions to increase worker safety or reduce consumption. What microsoft Y Goalthat They rely for this on Artificial Intelligence. But also Google.
All three are researching solutions to prevent and avoid security problems in data centers. For example, Microsoft is developing an AI system that is responsible for analyzing data from various sources, and generating alerts that it sends to the data center operations and construction teams. Your mission is “avoid or mitigate the impact of security incidents«.
In addition, those from Redmond are also developing a system, complementary to this and closely related to it, whose purpose is to detect and predict the impact that data center construction schedules may have. Of course, both initiatives, as they have told Techcrunch from the company itself, are still in the early stages of their testing phase, and their expansion to Microsoft production environments is not expected to begin until later in the year. .
As for Meta, he also claims to be investigating how Artificial Intelligence can anticipate the operation of your data centers under extreme environmental conditions, as these can lead to unsafe working environments. Until now, the Facebook parent company has been developing physical models that simulate extreme conditions, and passing the data obtained with them to the Artificial Intelligence models responsible for optimizing energy consumption, cooling and airflow in its servers.
According to comments from the company, they have a significant amount of information from their data centers. Specifically, they refer to the information they have from data center areas considered high frequency, which have sensors integrated into servers and racks. Also in the data rooms.
According to Meta, in the data centers, «each server and network device, with their different workloads, will consume different amounts of power, generate different amounts of heat, and emit different amounts of airflow«. Thus, they point out that their infrastructure team «it collects all the data from each server and then develops AI models that can place your servers and racks in data centers, and push workloads to these servers to optimize performance and efficiency.»
Apart from security, technology companies have other reasons to improve the operation of their data centers and make them always work optimally. Among them, the economic ones. If a data center experiences an outage, it costs the company money. And they are becoming more frequent, so they are interested in investing in systems that can predict these blackouts, and even avoid them.
Meta has more than 20 data centers around the world, with several new construction projects in the United States underway. Meanwhile, Microsoft has more than 200, and plans to build between 50 and 100 new data centers each year in the short and medium term. Hence the importance of AI being able to collaborate in keeping them online for as long as possible.
But in addition to this, Artificial Intelligence also promises to offer companies providing and managing data centers new opportunities in the field of energy management. This would lead them to be able to reduce costs, and therefore save on data center expenses. Back in 2018, Google noted that Artificial Intelligence systems developed by its subsidiary DeepMind they had reached get them to save 30% on energy costs in data centerscompared to your historical consumption data.
For now, DeepMind has no news or updates on this specific area apart from those initially discussed. But the use of Artificial Intelligence to save on energy consumption is not exclusive to Google. In fact, both Meta and Microsoft are already using it to save on energy costs. At the end of 2021, Microsoft launched anomaly detection methods powered by AI, whose mission was to measure, assess and mitigate unusual consumption of water and electricity in a data center.
To do this, these methods used telemetry data from electrical and mechanical devices. In addition, Microsoft uses Artificial Intelligence-based approaches to identify and repair faults with energy meters in these centers. Also to identify the most suitable points to locate the servers and to make them consume the least possible amount of energy, network and cooling capacity.
As for Meta, he says it has started to take advantage of reinforcement learning to reduce the amount of air it blows into data centers for cooling. Most of the company’s data centers use outside air and evaporative cooling systems, making airflow optimization a priority.
