How AI is transforming the data center

 In Data Center, Hardware

Data center growth is exploding. This growth is driven by the expansion of cloud providers, health care organizations, and financial service providers. Data-centric companies in retail, social media, and entertainment are harnessing the power of data to transform the customer experience. Smart cities are moving from vision to reality and generating massive amounts of data in the process.

Traditionally, data centers are designed around the confluence of large data sets, cheap electricity, and inexpensive land. This combination is driving about 80 percent of the world’s internet traffic through Ashburn, Virginia, the data center capital of the world. Dallas, New York, and Seattle are other growing data center hubs.

But the traditional data center approach is evolving, and data science and AI are now influencing the design and development of the modern data center. Artificial intelligence (AI) is driving new efficiencies that transform everything from the location of data centers to their silicon architecture in order to realize new applications.

Data center modernization being assisted by AI

Every industry is leveraging data in some respect to gain new insights and advance their field.

Health care and life sciences companies are examining genomics, biometrics, immunotherapies, brain initiatives, and so on to better predict ailments and improve therapies.

The transportation industry is using data to identify where accidents are likely to occur, eliminate rush-hour bottlenecks, and improve the safety of public transportation.

Law enforcement is combing through data to anticipate threats and improve public safety based on past incidents.

Now, more and more data centers are adopting AI as a way to modernize their operations. With AI, data centers can aggregate and analyze data quickly and generate productive outputs, which operators can use to manage density, reduce power consumption, and increase performance. Data centers operators utilize machine learning and AI in new and ingenious ways, ultimately driving efficiencies up and costs down.

Data science and AI build on the virtualized infrastructure that has enabled data center operators to add more workloads on the same physical silicon architecture. Data centers moving to hyperscale improve the density associated with compute, networking, and storage.

Ultimately, using AI, data center operators can optimize architectures for specific workloads. Whole buildings can focus on one workload, whether for health care, genomics, education, or weather. Two concepts are influencing that design.

How data science affects data center design

Data science is the first driver affecting data center design. Since data science is all about collecting larger and larger sets of data for analysis, designers and operators need to account for this when blueprinting data centers.

Data analytics capabilities work to align these large data sets with the appropriate amount of compute, power, and storage. By understanding the data lifecycle, designers and operators gain the ability to store more in data centers. As deployment of IoT devices connected with 5G and advanced networks increases, they will drive data to and from the edge, where it’s analyzed and delivered back to the data center.

Data center management and operations is the second driver. Google’s acquisition of DeepMind is a quintessential example of how AI enhances data center operations and management – after having some fun creating games and apps, Google put DeepMind’s AI to work monitoring servers’ air-conditioning units to prevent overheating.

It predicts how much energy the servers will expend for a specific function, and then tailors the air-conditioning usage to that demand – this made Google’s cooling units 40 percent more effective and slashed total electricity consumption by 15 percent.

That capacity and efficiency are leading to more optimized workloads. The same thing is happening at the level of integrated circuits, where data science helps accelerate performance to drive efficiencies.

Data centers are on the move

The trend of dedicating a whole data center to one workload is growing, but just as influential is the growth of mobile data centers.

One reason for this is that data centers are highly sensitive to the costs of electricity and real estate, and places with cheap land and power are becoming scarce.

Organizations are also investing in mobile data centers connected to 5G networks to improve disaster response times, increase safety measures, and avoid downtime.

Internet of Things (IoT) development is another big factor, driven by the need for edge capability. IoT doesn’t just demand that data centers be compact and mobile, it also requires this of the AI and data science itself. Now, machine learning algorithms and cognitive computing algorithms are compactly designed so they fit into mobile devices.

Companies are aiming to drive down the size and the computational and power requirements of these devices. As data science continues to drive data center growth, focus on mobile centers will only increase.

Where will AI take us?

Some might get hung up on the legal, ethical, and societal implications of AI – after all, data science is examining vast amounts of sometimes sensitive data. It’s always important to remember that the technology isn’t the goal; it’s the enabler.

Data science and AI transforms the data from exhaust we all leave behind into the fuel we use for the next great advancements in fields as diverse as medicine, cybersecurity, transportation, energy, climate, and more.

With more efficient, focused, and mobile data centers optimized with AI, those discoveries are even closer within reach.

About the author

Melvin Greer is Chief Data Scientist, Public Sector, Americas, Intel Corporation. He is responsible for building Intel’s data science platform through graph analytics, machine learning, and cognitive computing to accelerate transformation of data into a strategic asset for Public Sector and commercial enterprises.

Related Posts: You may also be interested in...


Leave a Comment

19 + 15 =

Pin It on Pinterest