Most IT leaders like to believe they have a complete grasp on data center management, operation and planning. Actually, they don’t.
There’s no way any IT leader or team of IT experts can exert second-by-second (or even more granular) control over essential data center tasks. Humans — even highly educated and trained ones — also tend to allow personal preferences, prejudices and misconceptions to cloud their views on future planning and other critical responsibilities.
Artificial intelligence (AI) has none of these shortcomings. That’s why, even as data center operators fret over hybrid environments, the internet of things and other challenges, they also need to consider the impact AI is beginning to exert on an array of key data center operations and services.
Here’s a look at seven things every IT leader needs to know about how AI is transforming the data center into a more powerful and efficient facility.
1. Many different types of data centers can benefit from using AI
Any type of data center can benefit from AI, but the ones that generally benefit the most are large-scale facilities, such as big enterprise data centers, public cloud data centers, co-location data centers and outsourced data centers, says Joe Merces, CEO of enterprise backup and disaster recovery technology provider Cloud Daddy and former CIO of the New York City Law Department.
Tom Coughlin, an IEEE Fellow and president of Coughlin Associates, a data storage analysis firm, believes that any data center can take advantage of AI methods, such as machine learning, to better manage internal resources, as well as to anticipate upcoming hardware and data requirements. “AI is becoming one of the most important [data center] applications,” he notes.
Machine learning is moving from basic pattern recognition and traditional algorithms into the more sophisticated area of deep learning, explains Paul Mercina, innovation head at data center maintenance service provider Park Place Technologies. “A key contribution of machine learning is its ability to discover structure within data using an iterative approach, without needing humans to begin with any theories or assumptions to test,” he says. Deep learning uses multi-layered artificial neural networks to deliver high-level accuracy in tasks such as object detection and classification, speech recognition and language translation.
2. AI helps data centers become more energy efficient
Over the past few years, AI tools have played an increasingly important role in reducing data center energy consumption and waste. “These applications help reduce power, report cooling inefficiencies and analyze the health status of critical mission systems to improve efficiency while saving energy,” Mercina notes.
“A data center is an ever-changing environment,” observes Stijn Grove, managing director of Dutch Data Center Association. AI-powered analysis and monitoring of current indoor and outdoor temperatures, as well as predicting upcoming weather conditions, enables data centers to optimize cooling resources and conserve energy, he advises.
Servers are any data center’s hungriest energy users. “When you can automatically scale cloud servers up or down when needed, use each server to the fullest [potential] and shut down not-used capacity, you win a lot,” Grove notes.
AI can also significantly trim storage energy consumption. By using AI monitoring and analytics to anticipate various types of user activities, data centers can quickly shift less frequently used data to lower energy storage resources and move more frequently used data to higher performance storage. “Also, it may be possible to use AI to minimize the movement of data back and forth during processing,” Coughlin says. “Smart data placement for data in active use can position data and processing closer to each other, thereby reducing energy expended by excessive data movement,” he explains.
3. AI leads to stronger data center security
Data center security needs are evolving rapidly. Until recently, the biggest threats data centers faced came from employees or relatively primitive external brute force attacks. “Now, hackers are building AI-based algorithms, which are trying to find weaknesses in a data center,” reports Param Vir Singh, associate professor of business technologies at Carnegie Mellon University’s Tepper School of Business. AI is the technology best positioned to handle this challenge, he notes.
“AI applications are enabling data centers to adapt more quickly to ever-changing security requirements while also providing a more secure environment for their users without forcing strict rules,” Mercina says. “AI solutions can also assist with detecting malware and spam, analyzing normal and abnormal activity patterns, identifying weak spots and strengthening protection from potential threats.”
AI can also be used to trap malicious intrudes into ‘honeypots’ “where they can be monitored more closely and even be used to trace back the intruder,” Coughlin observes.
4. AI can optimize data center performance
By relentlessly monitoring and tweaking resources, including processing, networking and memory, AI can allow enterprises to run their data center at maximum efficiency. “AI can be used to monitor workload distribution, making infrastructure more scalable as well as optimizing performance for efficiencies in cooling and power consumption,” Merces notes. AI can also be used to optimize server configuration and utilization. “For example, AI can recognize infrastructure problems and self-heal by moving workloads and attempting repairs through restarting, power cycling and re-imaging,” he says.
AI is uniquely effective at optimizing server use, Coughlin observes. “This can include moving appropriate processing to application-specific processors, such as GPUs and TPUs.” AI can also optimize data center software performance. “For instance, to limit the polling of the same data in databases or limit duplicate processes,” Grove adds.
5. AI is set to improve infrastructure management
According to a Ponemon Institute study, the average cost of data center downtime across industries was approximately $8,850 per minute in 2016. “If we can predict a maintenance issue even before it occurs, we can take preventive action,” Singh says.
Using ever-improving infrastructure management technology and smart sensors, neural networks can be trained to analyze the existing demands and capacity of infrastructure to tap the most appropriate equipment to meet the demand. “Because AI can process dramatically more information than a human or a team of humans, near instantaneously, an AI-directed system is more efficient and more reliable,” says Tod Northman, a partner at law firm Tucker Ellis, specializing in business and corporate law. He notes that sensors can also help data center managers predict or mitigate catastrophic failures.
Today, most data centers are managed, monitored and serviced by highly trained individuals executing mundane tasks, such as traversing data center rows, searching for lights that indicate hardware failures, Mercina notes. “AI and machine learning have the power to completely transform this outdated paradigm by stripping away the guesswork and overlaying proactivity to the entire ecosystem.”
AI promises to have a huge impact on how data centers schedule routine maintenance tasks. By scrutinizing all relevant data center resources, AI will soon be able to predict, to the moment, when specific facilities will need service, upgrading and replacing. As a result, scheduled maintenance program will gradually be replaced by AI-generated recommendations, Grove predicts. “This will improve uptime and lower costs,” he reports.
6. AI is becoming a powerful data center planning tool
Planning is one of the most exciting AI data center applications. By drawing massive amounts of information from data center sensors, and using its ability to learn from previous situations, AI can provide granular forecasts and, more important, model the differences in changed assumptions, Northman says. “The longer the system is in place, the more information the system will have and the better its forecasting will become.”
“It’s happening today,” Merces reports. AI is being used to plan and provision power resources, for example, as well as to forecast cooling needs. “It’s also being used to plan and manage network and bandwidth utilization and optimization,” he notes.
7. AI will manage an increasing number of data center tasks with little or no human involvement
An almost total encroachment by AI on data center tasks currently handled by humans is very likely, Grove says. “The digital ecosystem demands more instant controls and operations that can be only achieved by AI and machine learning,” he states. “Also, with the emergence of edge computing, to be able to manage many unmanned sites you need AI to do it properly.”
The holy grail is a fully automated data center that monitors, diagnoses and heals itself. “This requires AI, robotics and even augmented reality technologies — machines taking care of each other,” says Roger Brooks, chief scientist at Guavus, a big data analytics firm.
On the bright side, at least from a human standpoint, is that fact that AI still isn’t close to performing high-level reasoning and decision-making tasks with any degree of reliability. “With all the work going into AI, it’s all being zoned-in on specific functions, which will become as efficient as they can possibly get but, ultimately, they will not become intelligent,” Merces predicts.
Northman agrees. “While managers will increasingly rely on AI to operate and manage data centers, I do not foresee entirely removing humans from the process,” he states. “Managers’ roles in particular centers will be reduced … but humans will remain in the loop as a fail-safe.”