What does artificial intelligence mean for your data centre?

Klaus Manhart, Computerwoche

Monday 10 April 2017

Artificial Intelligence has finally arrived. After years of stagnation and research, the discipline is now experiencing a true boom phase with far-reaching consequences and opportunities for companies. But what does this mean for your data centre?

Artificial Intelligence (AI) has the potential to fundamentally change IT. It’s now possible to simulate learning, drawing conclusions and natural human communication with the help of new software. AI systems answer questions in a natural language, independently search through databases and give answers in the shortest amount of time. The results are impressive and generate true business benefits.

AI systems are now being installed on even the smallest mobile devices. Smartphones equipped with speech or facial recognition features are primarily based on machine learning algorithms. And in individualised online advertising, we have touchpoints with machine learning algorithms that are not immediately apparent.

However, AI is made up of more than just learning algorithms. It also involves diverse specialised fields and methods. These include knowledge-based systems as well as the fields of robotics and pattern recognition. Machine learning and, especially, deep learning processes are currently considered the central and most successful AI method.

The Lenovo Smart Assistant is a good example of deep learning processes, as is object recognition in photos. In the meantime, deep learning has become one of the machine learning processes that major IT companies prefer to use.

AI has enormous potential

Analysts attest to the great potential of AI when it comes to its business applications and improving a company’s competitive position. “In the next 20 years, artificial intelligence will change our economy and how we work more than any other technology in the recent past,” says Dr. Matthias Kaper, head of Artificial Intelligence in the Emerging Technologies Division at Accenture. “Even in the service industry, there are numerous deployment options for artificial intelligence, whether as virtual agents answering customer queries or automating workflow sequences that require documentation.”

According to one study, conducted by the market research company Crisp Research, 64% of IT decision-makers are dealing with machine learning and other AI technologies, and some 20% are even using machine learning productively.

The most important motive is to boost customer loyalty. AI does this by streamlining communication with customers using digital and natural-sounding assistants and assessing potential user behaviour.

AI systems can not only be used in customer service, but also in industry, and are increasingly being used for digital transformation, from intelligently controlling production equipment to monitoring connected devices.

Hyperconverged systems

In addition to theoretical, technological and software advances, powerful computers and modern hardware are able now to rapidly process large amounts of data, making AI systems useful in many different applications. The first thing companies should do to ready their IT for AI is update their data centres.

Hyperconverged systems are especially suitable for this purpose. They are the infrastructure-based answer to AI and machine learning because they enable local on-premise systems, external data centres and cloud services to be seamlessly and securely integrated.

This enables them to flexibly scale storage and computing power so they can be adapted to changing needs at any time. It also enables businesses to move large amounts of data to the cloud and, when needed, transfer it back and forth between local and remote systems.

Lenovo’s hyperconverged systems support companies with AI ambitions. Five computers from the Lenovo Converged HX series, equipped with the Nutanix Xtreme Computing Platform (XCP) software, are currently available on the market and are ideal for AI purposes.

The HX series integrates software from the leading brand for hyperconvergence in the extraordinarily reliable and scalable servers from Lenovo. The flexible modules consist of completely integrated and tested computing and storage resources, and pre-installed virtualisation management software, which makes simple scale-out-clusters possible.

High speed is the trump card

When it comes to performance and speed, though, as mentioned earlier, powerful hardware is a requirement. Today, the way that most AI applications get processed is simultaneously and via a hardware platform that uses numerous graphic processors (GPUs). The reason is that GPUs have many computing units, enabling them to anticipate considerably higher computing speeds compared to solutions that work exclusively on CPUs.

In the coming years, AI applications are going to significantly change service industries, especially customer service, but also industrial processes. With the right technologies and strategies, companies can gain competitive advantages.

To do so, however, they need to have IT architectures in place that provide the necessary performance, speed and resources for processing huge amounts of data. Hyperconverged infrastructures, like those offered by Lenovo, can effectively assist businesses in meeting these requirements. And they achieve this by seamlessly and securely integrating local on-premise systems, external data centres and cloud services.

YOU MIGHT ALSO LIKE

Building the next-gen data centre

Where traditional and web-scale apps co-exist