Everest 5G Engineering Services PEAK Matrix® Assessment
NTT DATA is recognized as a Leader and Star Performer. We support enterprises with our innovative solutions and 5G engineering capabilities.Get the Everest 5G report
How to Choose the Right Approach for Application Modernization and Cloud Migration
Rehost, replatform, rearchitect, rebuild or replace? Use the Gartner evaluation framework to choose your approach.Get the Gartner report
Topics in this article
It’s anyone’s best guess what the technology landscape will look like 15 years in the future. As AI, machine learning and the Internet of Things continue to evolve at a lightning pace, it seems like literally anything is possible. But what about the evolution of the data center?
There’s a misconception that data centers are monolithic entities that still act solely as data warehouses. Nothing could be further from the truth. Modern data centers adopt the latest technologies to keep pace with the ever-changing dynamics of the data explosion.
Let’s look at what a data center might look like in 2035.
The green revolution
There is a vast customer-driven initiative toward greener data center practices like reducing carbon footprint and more energy-efficient cooling solutions. Since a great data center is inherently customer-centric, they are listening and responding in a variety of ways.
By 2035, we’ll see the seeds of the green revolution bear fruit in a variety of ways. Immersive cooling, where servers are literally immersed in water for cooling, may gain enough inroads to make them ecologically friendlier solutions than traditional cool air. Another possibility is already taking off: blowing cool air directly through the server components. This makes it possible to cool servers in a much more compact and direct way, which should in turn make the process greener and less power-hungry.
We can also look to the supply chain to further greener goals. Ultimately, the greater costs associated with more ecologically friendly operations will need to be passed along, and customers and procurement will need to cooperate to reduce the impact.
Since data centers are the heavy factories of the IT world, it’s always important to evaluate each step in the process to ensure that performance and reliability don’t falter. That said, new technologies offer an exciting future for data centers that isn’t quite possible or feasible today.
Lights out for the data center?
Lights out management (LOM) is a hot topic in the data center community today. With all the trends toward automation, it may seem like in 15 years robots will be running the data center. But is that necessarily where the trend is taking us?
The stakes are too high for traditional data centers, and the SLAs they have in place demand a core team present 24/7 to keep up with 24/7 uptime. Response times are shrinking, so there is really no sound way to automate for every contingency that can arise.
Methods for reducing the number of personnel required to handle crisis incident management are on the horizon, but emergency response times must remain extremely fast. There’s no way to automate crisis management effectively, and the traditional data center will never be LOM. The people involved in managing essential services will always be cost efficient in the end.
The growing sophistication of AI
How will data centers keep up with the exponential growth in machine-to-machine communication? Simply put, AI drives the scaling up of the data center. The data center business remains a cautious and conservative one, even though data centers house some of the world's most sophisticated technologies.
While AI is not yet a mature part of the data center industry, a great deal can change in 15 years. What is certain is that the industry's inherent caution will ensure that any future AI solutions will be thoroughly vetted and evaluated. Will the solution end up increasing cost while increasing reliability? That’s the balance data centers need to strike – profitability and progress for customers.
Every data center will tell customers they provide five-nine uptime, but a great data center strives for 100%. Being conservative with AI will help maintain the balance while increasing data center efficiency and resiliency. The need will drive the AI solutions, not the other way around.
The future is software-defined
As software-defined infrastructure continues to make rapid inroads into enterprise computing, how will that model affect the data center’s traditional function? For one, virtualization has changed the patterns of usage. These patterns were slow to change, but now with technology-assisted virtualization the data center is more dynamic.
As we progress toward 2035, data centers will need to make sure they’re able to support the levels of power they’re providing to customers. The best data centers never fail to provide what they promise in terms of power and cooling, as well as promises to maintain the infrastructure.
We can expect customers to ramp up toward software-defined infrastructure. Software-defined processes are only now creeping out of infancy into viable applications—provisioning power virtually being a good example. This will accelerate in the next 15 years, and data centers will be designed and built to be more customer-focused rather than capital-focused.
What other changes can we expect to see? Increasingly, distributed computing is on the rise, and we can see that trend continuing. We may get to the point where every device has its own distributed micro data center. How the data center industry will respond depends on the evolution of distributed computing technology, but you can rest assured that data centers will always be on the forefront of advances.
15 years is forever in technology. While we can speculate on the effect that new and emerging technologies will have on the data center, the future holds endless opportunities. One thing is for certain, though. The data center industry will grow exponentially, and hopefully, sustainably. But whatever the actual data center of 2035 looks like, I have a feeling it's going to be exciting.