Event Recap
RECAP | ROUNDTABLE | AI in Data Centers: Today and Beyond
26 Jul 2018
Uptime Institute’s Rhonda Ascierto and Andy Lawrence led a discussion focused on artificial intelligence (AI) in data centers. Online roundtable participants shared their early-stage and high-level plans for AI, which mostly focused on operational use, including efficiency improvements.
One participant is looking to leverage DCIM data for AI, while another was exploring AI as a potential offshoot of their standardization approach using BIM (building information modeling). BIM will support their efforts in identifying sites, getting accurate costing, speeding up designs, and so on, which it may be able to integrate operational data into, and leverage using AI, at some future point.
This fed into a discussion about what AI really is in the data center, which led to the topic of how people were using AI. To date, it has mostly been used as part of data center cooling controls (via software such as Vigilent and SynapSense [now part of Panduit]. It was noted that AI in data centers today is mostly being used to improve existing functions but that over time new functions are expected to emerge.
All participants were still in the early stages of exploring AI. There was a full discussion on how difficult it can be to get started, including the ability to amass enough data. One participant had a large-enough facility portfolio to likely enable them to run AI models using just their own data, while others are thinking about DMaaS (cloud) services, in which their data would be anonymized and pooled with others, with analysis specific to their data center(s) sent back to them.
The availability of internal vs external skill sets (data scientists) to support AI also came up, as well as data ownership, and liability concerns in the event of an incident that can be related to AI-driven analysis or automation.
In closing, there was agreement that AI use cases and case studies, as well as more research into the legal ownership and liability of AI-related data, would be beneficial for the Network and the industry as a whole.
One participant is looking to leverage DCIM data for AI, while another was exploring AI as a potential offshoot of their standardization approach using BIM (building information modeling). BIM will support their efforts in identifying sites, getting accurate costing, speeding up designs, and so on, which it may be able to integrate operational data into, and leverage using AI, at some future point.
This fed into a discussion about what AI really is in the data center, which led to the topic of how people were using AI. To date, it has mostly been used as part of data center cooling controls (via software such as Vigilent and SynapSense [now part of Panduit]. It was noted that AI in data centers today is mostly being used to improve existing functions but that over time new functions are expected to emerge.
All participants were still in the early stages of exploring AI. There was a full discussion on how difficult it can be to get started, including the ability to amass enough data. One participant had a large-enough facility portfolio to likely enable them to run AI models using just their own data, while others are thinking about DMaaS (cloud) services, in which their data would be anonymized and pooled with others, with analysis specific to their data center(s) sent back to them.
The availability of internal vs external skill sets (data scientists) to support AI also came up, as well as data ownership, and liability concerns in the event of an incident that can be related to AI-driven analysis or automation.
In closing, there was agreement that AI use cases and case studies, as well as more research into the legal ownership and liability of AI-related data, would be beneficial for the Network and the industry as a whole.
Request an evaluation to view this report
Apply for a four-week evaluation of Uptime Intelligence; the leading source of research, insight and data-driven analysis focused on digital infrastructure.
Already have access? Log in here
Posting comments is not available for Network Guests