WavebreakmediaMicro - Fotolia
As the IT market evolves, emerging technologies, such as artificial intelligence and machine learning, can have a significant effect on the day-to-day operations of a data center. Data center administrators will most likely face these new technologies in two ways: by making room for data-heavy artificial intelligence applications and by using data center tools with these capabilities. SearchDataCenter's advisory board predicts the future of machine learning and artificial intelligence in data centers and how you should prepare your IT team for the changes.
Robert Crawford, systems programmer
Artificial intelligence (AI) and machine learning (ML) have a lot of promise, especially for companies wanting to anticipate and seamlessly fulfill customers' needs. AI and ML also hold a lot of promise operationally. I hope it brings us closer to self-healing systems able to anticipate, route around and fix problems before they become serious. They also may be useful for predicting and allocating computer resources without human intervention. IBM has already done some of this with zAware and Workload Manager.
This year, IBM announced Machine Learning for z/OS, which seeks to make machine knowledge available to mainframe apps. IBM has tried to further strengthen its position as a leader in ML by offering zAware, which basically learns how a mainframe shop works so it can flag anomalies.
Having said that, the current hype about AI and ML reminds me of when the World Wide Web first hit home desktops in the '90s. Some pundits declared the impending deaths of "brick and mortar" and mainframes. Only now, 20 years later, are these predictions remotely becoming true. Therefore, I suggest we take a deep breath, ignore the hype and work on what's possible.
Lex Coors, chief data center technology and engineering officer, Interxion
I believe that it is now the time to make a strong disconnect between ML and AI to avoid a lot of confusion in the years to come. ML will continue to develop with algorithms where, without emotion and following calculated rules, decisions will be made on the best result. Artificial intelligence in data centers can make decisions based on the same rules as ML plus taking emotions, intuitions and feelings into the equation. The difference is that AI allows for different ways of thinking.
ML has already begun to affect data centers, where thought leaders, like Schneider Electric, have incorporated analytics based on ML that will help to manage our ever more complex and growing data centers.
Large amounts of data, which previously led to additional full-time employees who tried to make sense of it all, are now, via analytics, converting into meaningful insights. These insights can decrease employee workloads, improve data center efficiency and decrease operational expenditures. For example, organizations relied on planned maintenance in the past, but now, there is more dynamic maintenance -- or maintenance when we need it. Eventually, ML can also lead to a decrease in total cost of ownership, enabling enterprises to identify the performance of individual system components in relation to the whole infrastructure.
Clive Longbottom, co-founder, Quocirca
AI and ML have the capability to transform how data centers are run. Increasingly, the move toward virtualized and cloud-based platforms means that administrators are struggling to deal with issues. For example, the root cause could be down to any of several items, and the curing of one issue can just shift a problem to a different area on the platform. AI and ML can work from the creation of a known baseline operating condition for a platform and can then monitor any significant change to this -- and can see exactly what has caused that change. Based on empirical knowledge of the multiple different workloads concerned, the system can then automatically decide whether the change requires intervention or whether it should just wait to see if the event calms down and the platform reverts to normal conditions.
Also, AI and ML can more intelligently deal with major changes to the network. Such intelligence can embrace new resources as they become available, doling them out as necessary. AI and ML engines can more easily deal with the failure of any item within the platform. Machine learning and artificial intelligence in data centers can analyze unexpected workload peaks in real time and make decisions as to whether this is technically anomalous, like a resource leak; maliciously anomalous, like a distributed denial-of-service or intrusion attack; or just caused by a peak workload, such as a marketing campaign initiated from within the business.
Overall, IT managers should be embracing machine learning and artificial intelligence in data centers as rapidly as they can -- as long as they carry out due diligence as to the reality of the intelligence of the chosen platform.
Prep your data center for AI apps
Experts review top data center challenges
How AI can improve data center storage