Against the background of the seemingly inevitable total transition to artificial intelligence (AI) and automation technologies, Leo Craig, CEO of Riello UPS, asks: is the uprising of machines inevitable and, if so, how will this affect the data centers in the future?Almost a third of the UK population uses at least five plug-in devices every day. This is approximately 20 million people with more than 100 million smartphones, wearable devices, virtual assistants and other gadgets. And this is just the tip of the iceberg. Experts predict that by 2020 more than 50 billion devices will be connected to the Internet of Things (IoT) around the world, and according to authoritative International Data Corporation, the emergence of a 5G wireless network over the next decade will create a market worth $ 7 trillion a year.
There are already more connected devices on the planet than people. And this means only one thing - a lot of data, a lot of data to process! In order to keep up with the ever-growing demand, modern data centers should do more and more, using less and less work space, energy, and ideally, human resources.
The only way to solve this complex task is to carry out a technological revolution, known as Industry 4.0. Automation, big data, artificial intelligence, machine learning - the “smart” data center is our only way out.
So, cars are coming ... But will they manage to take complete control?
What is data center automation?
In simple terms, this means automating the entire spectrum of tasks that people typically perform: from managing and controlling data to maintenance.
The main goal: for tasks that require decision-making, use not people, but machines that are able to “learn” and make such decisions by analyzing information from sensors on connected devices. Predicting future results based on past data.
A modern data center has become so large and complex that people often do not have time to process the entire stream of various information.
Machines, in contrast, can analyze and process data almost as quickly as they receive it. With smart plug-in machines, this means instant real-time response. If necessary, managers of the data center can perform an instant scaling of the resources used, which will ensure the rapid processing of many errors without human intervention.
In addition, the cars do not sleep, they do not need a vacation and sick leave: they are at work 24 hours 7 days a week.
Machine-to-machine communication allows the automation of many operations performed by IT administrators and information technology service groups using AI technology. An uncountable amount of raw data in an instant becomes comprehensible and practically applicable.
In conjunction with data center infrastructure management software (DCIM), AI can automate many key tasks, such as isolating security threats by remotely disconnecting third-party systems.
Take on the time-consuming responsibilities of patch management, updating systems to the latest, safest versions. AI can be used to perform all tasks: from launching disaster recovery processes to the relatively prosaic but important routine of automatically generating daily reports.
Of course, this kind of “smart” communication is not an invention of today. One of the striking examples of the successful application of this technology is an uninterruptible power supply. The sensors of the UPS collect and transmit to each other information both about the indicators of the device itself, for example, the battery charge, and about the external environment.
Modern UPSs can also communicate and integrate with “smart” power grids, which allows electricity to be stored, intelligently consumed and controlled for its demand. Demand management, known as DSR, is a concept in which energy consumers are provided with incentives to reduce consumption during peak periods, which helps the power grid balance demand and supply without generating additional energy.
Thus, in practice, data centers can use their UPS to store excess electricity during periods of off-peak load when the unit cost is lower, which can then either be used during peak loads or power cuts, or sold back to the supplier. Although to operate in the “virtual power station” mode, the UPS should be equipped with lithium-ion batteries, but in the end it will allow the company to declare itself as a socially responsible and environmentally-friendly organization, as well as provide an additional source of income.
So, why is the introduction of automation in the data center beneficial, in addition to the obvious increase in free time of the IT-team?The clear advantage is speed and guaranteed expected performance with minimal need for human intervention. Potential problems can be identified and resolved in real time, thereby increasing the productive time of the data center and reducing the harmful periods of downtime. Another area with great potential is to increase efficiency and reduce energy consumption of the data center.
AI in action: DeepMind in service with Google
One of the most striking examples of the incredible results that can result from the introduction of automation in a data center is Google. Back in 2014, the technology giant began using machine learning from the recently acquired AI unit, DeepMind, to manage energy consumption in its extensive network of data centers.
After analyzing the historical data from
120 variables , such as energy consumption, pump speeds and temperatures, the AI algorithms were able to calculate and create much more efficient ways of cooling data center equipment.
What is the result? Cooling demand decreased by
40% , and total energy consumption by
15% - an impressive result, which saved the company hundreds of millions of dollars.
Although it is obvious that most other companies do not even have such resources close to them, and often technical knowledge like Google, moving in this direction can make a significant contribution to optimizing the work of data centers, regardless of their size.
Indeed, a modern UPS connected to the network can be a great help in “smart” control and power consumption in such a usage scenario. UPS sensors continuously collect statistics on key performance indicators, based on which, as well as on the analysis of the quality of the mains supply, the device can go into power saving mode if necessary. As a result, the efficiency of the UPS increases
to 99% . This optimization allows data centers to save energy and reduce carbon emissions.
Brave new world?
Now, in most cases, the automation of processes in data centers comes down to providing information to its employees, people, in an easy-to-interpret way: the main share of the decisions taken is still left to people. However, with the improvement of AI technologies and machine learning, this separation of powers will become less and less obvious.
If previously data centers could only serve as a means of storing data, then the new “smart” data centers of the future will already be able to analyze, interpret and process all incoming information in real time.
Rise of the machines can significantly affect the operation of data centersAt all times, the introduction of new technologies and technological processes was greeted with caution, as this foreshadows the reduction of jobs: cars take on the tasks previously performed by people. For the first time, these fears were voiced by Luddites, and since then each new stage of the industrial revolution has met with such resistance.
But if we carefully study history, we will see that only the nature of the person’s work actually changes, the workplaces themselves do not become less. And, probably, this trend will continue in the case of "smart" data centers.
Compared to many industries, the number of people required to manage a data center is already quite limited. Perhaps the spread of AI will lead to a reduction in the number of positions of IT-specialists with a shift towards experts of a wider or universal profile.
Automation should be considered by managers and administration of data centers as an additional tool in their arsenal, and not cause fear of change. It will allow IT-specialists to concentrate on tasks that bring practical benefits, and not to engage in "maintaining the engine and car in good condition," if you draw a analogy with driving.
In the end, it does not matter how perfect AIs and machines are: they are only as good as their algorithms and programs are thought out. Machines are vulnerable to the effects of people or even other machines, besides, no one has canceled data privacy and data security issues.
Despite the fact that AI, automation and machine learning technologies will definitely play an increasingly important role in all aspects of our life, we are still far from the world (or in our case, data centers), which is controlled exclusively by robots.