Data centers stand at the forefront of IT infrastructure strategies and innovation. Far from a mere warehouse filled with servers, the modern data center is a technological marvel, powered by cutting-edge equipment and micromanaged down to the last detail by powerful software applications.
Despite all those resources, most data centers still rely upon well-trained, experienced IT professionals to oversee operations. But for how long? Advancements in automated processes and machine learning have led industry experts to wonder when completely automated data centers will become not just common, but standard practice.
With advanced analytics and artificial intelligence tools at their disposal, data centers are already implementing automated processes to handle low-level tasks that were once carried out by human technicians. In addition to offering better efficiency and performance, automating low value work frees up time for IT personnel to focus on more mission critical operations and provide higher level services to clients.
Here are a few examples of tasks that are currently undergoing automation in a new generation of smart data centers:
Many routine data center functions that are vital to keeping the facility running smoothly are rather easy to perform with minimal effort. Data backup, replication, and other application events are just a few processes that can easily be set up as automated tasks and scheduled to be performed regularly. Other repetitive and time-consuming tasks like implementing patches and setting configurations can be addressed through automation.
One of the most time-consuming tasks for any IT department is dealing with tickets generated by incidents throughout the systems being monitored. In many cases, these tickets don’t represent serious issues, but rather temporary errors or minor problems that are easily addressed. Rather than taking up hours of a system administrator’s time, the process of filing and reviewing tickets can be handed over to automated tools that can deal with minor problems while still passing more serious concerns on to human technicians.
Data centers are already using predictive analytics to improve efficiencies, minimize waste, and reduce costs. Statistical algorithms and AI-driven machine learning can monitor every system in a data center and identify potential problems before they develop. When simulations and analysis in a smart data center find that a system is likely to fail within the next few months, it can be replaced more quickly and efficiently, rather than suddenly failing and compromising data center performance.
A data center is a complete environment, with multiple systems working in concert to deliver services. Power and cooling requirements can fluctuate rapidly based on usage needs. Automated tasks and systems informed by predictive analytics that monitor usage trends over time and model future demands for optimal power consumption, which can reduce energy costs and improve overall system reliability.
In 2016, researchers at the University of Pisa published a paper arguing that a future-world filled with hyperscale data centers will need to be fully automated if facilities are to have any hope of managing the huge number of day-to-day issues their servers generate. Their automated data center solution involves the development of machine learning systems that analyze data center logs and event records to learn how to regulate and respond to system changes in real time. They describe the approach as “Autonomics 2.0.”
That same machine learning could also be used to greatly reduce the amount of data being stored. With data centers facing intense pressure to meet the world’s ever-increasing data demands, any strategy capable of improving storage efficiency will certainly get plenty of investment and attention in the coming years.
Of course, the idea of the completely automated, or lights out, data center is hardly new. AOL launched a completely unmanned facility in 2011, largely as part of an effort to locate smaller data centers closer to end users on the outer edge of their network (a strategy known as edge computing today). When integrated with other smart data centers as part of a larger network, unmanned facilities can shift into a maintenance mode when something goes wrong and offload processes to other data centers until technicians can address problems.
While cost is frequently cited as the main reason for the push toward automation, data centers in rural markets are often confronted with talent shortages that make automating processes necessary. Automated data centers powered by cutting edge AI could be situated in any location, regardless of whether qualified technicians can be found locally. For companies implementing edge computing architectures, these unmanned facilities could greatly expand network reach and performance.
While total automation may not be on the horizon for every data center, the benefits of partial automation are already being realized by leading data center providers. As the machine learning technology that makes smart data centers possible becomes more commonplace and affordable, unmanned facilities will likely become more attractive for companies looking to expand their network reach with a limited infrastructure footprint. For the foreseeable future, data centers may still rely upon some degree of human management, but rapid developments in AI leave it an open question as to how long that future will last.