Artificial Intelligence is no longer the stuff of science fiction and movies such as "Minority Report." It's here and it's being put to practical use in data centers across the world. Learn how Google's AI experiements could result in lowered costs and increased efficiency in your data center one day.
Image: Flickr Creative Commons
Data Centers and the Artifical Intelligence Movement
Artificial Intelligence (AI) dominates the technology conversation today. AI has been an ambitious goal since 1956 when it was first defined at a workshop at Dartmouth College. The original objective of AI was to create machines that could simulate human learning and intelligence. Since then, this broad definition has become more structured. Today AI can mean:
- Neural Networks — Complex learning and used for higher level problems.
- Machine Learning — Learning through gathered information.
- Deep Learning — Learning through data analysis and data abstractions.
Most of the recent AI revolution has been based on the progress of machine learning and deep learning. It has wide-ranging, practical uses for different industries. The self-driving car, digital assistants, face-recognizing cameras, smart power grids, precision medical diagnosis tools and more are direct results of the recent innovations in AI.
Data centers have also started taking advantage of the AI movement. For starters, it is helping data centers improve efficiency while lowering costs. But there are other possibilities on the horizon too.
The World of Data Centers and AI Collide
AI is bringing new ways to help businesses get the most out of one of their most expensive investments - the data centers. The early focus is on the low-hanging fruits.
Last year, Google declared that it reduced electricity usage for data center cooling systems by 40 percent using deep learning technology. The deep learning AI gathered information from sensors on the server racks and then used regulated water pipes to cool the data centers. AI enabled Google to achieve this high level of energy efficiency. This has served as an inspirational story for many companies, with quite a few businesses working to improve data center energy usage through AI deployment.
AI based energy control proves that there are other opportunities. AI can be used to optimize server storage systems, support predictive analytics to find possible fault points, improve processing times and reduce risk factors. Anything that produces data can be fed into machine learning and deep learning tools to optimize for performance. Data center loads can become more predictable and more easily manageable.
Security and Maintenance
AI is helping automate cars, medical equipment and day-to-day human tasks. Similarly, it can help automate data center security and maintenance tasks in the future. When there is a security attack today, data center workers have to secure the servers manually. Also, a lot of maintenance work, like getting rid of faulty servers or storage devices, are still manual tasks. Adding robotics to the mix, data centers can become self-reliant and human-free. In the future, AI will be able to guide robots to perform the maintenance work. It will mean higher Quality of Service (QoS) at lower operational cost.
Google’s success in lowering costs of its data centers through the use of deep learning has inspired a lot of entrepreneurial companies. These companies are pursuing AI-based data center solutions and it seems likely this trend will continue. There are serious financial incentives for AI service providers. With the introduction of more AI-based solutions, data centers will be able to take advantage of decreased operational costs and increased productivity. It is a win-win for both AI service providers and data center owners.