Event Recap
RECAP | ROUNDTABLE | Data Center Staffing in the Age of Automation
As data center global capacity continues to expand, the availability of specialist staff will be an increased concern. Uptime Institute projects global data center staff requirements will grow from about 2.0 million full-time employee equivalents (FTEs) to nearly 2.3 million in 2025. In the U.S. and Western Europe, there is concern that many employees are due to retire around the same time causing an additional surge in demand. Automation is being touted as a way to help offset this staffing need, but innovation in artificial intelligence (AI) and automation are unlikely to have a major impact before 2025.
In this roundtable, the focus was on what members are presently doing in the New Normal, and what they plan on doing around data center staffing. Have you changed your staffing model to reflect the New Normal? Do you have a staffing plan and does it address staff availability through 2025 and beyond? Are you presently or planning on using automation and AI as a way to help offset staffing needs? What other staffing innovations are being considered? Rich Van Loo, VP Operations Services for Uptime Institute attended to add his experiences and insight.
Roundtable comments (presented in the form of a conversation with attendees):
Attendees were across 4 major verticals (financial, retail, telecommunications, and colocation) and provided their insights and comments on their present onsite staffing plans and the role automation might play.
Attendee 1 (A1): Presently operating fully staffed, 24x7. However, we are seeing a bit of staffing issues because of our close proximity to other data center sites in the area. Present concern over using automation and losing staff.
Attendee 2 (A2): Trying to figure out where we’ll be 5 years from now. Big part of evaluation is how automation will impact staffing. Also do we stick with the specialist model we presently deploy? Can we shift to more generalists with automation? We’re having issues finding specialists.
Attendee 3 (A3): Interested to hear about automation, AI, machine learning. Sense is we’re on the bleeding edge and not really seeing some of these things applied. Are others further down the line implementing this type of approach?
Attendee 4 (A4): Presently teaches introduction to data center operations at Northern Virginia Community College. Emphasis on entry level type of folks to fill staffing needs.
Rich Van Loo (RVL): Clients just starting to look at automation as a way to reduce onsite staffing, having people on-call, but not yet implementing for the most part. Sites are implementing more detailed monitoring at the sites as a part of this.
A2: We’re fully automated today with BAS/EPMS/SCADA/etc., with full monitoring that goes to onsite staffing. Presently down to single technician specialist during off hours. Our present goal is 2 technicians Monday-Friday, 24x5. All systems are remotely accessible now as well but requires onsite staffing to interact. We do want to get where we can allow remote interaction.
A3: We’re keenly aware of the shortage of qualified personnel. That of itself, however, is not driving staffing decisions. Confidence this issue of lack of staffing is being addressed and the pipeline will start filling. We are looking at if we have to have reduced staffing, how we deal with that during a sustained period to operate remotely – a lot of attention into that. We’re open to understanding how technology can help.
Question was then asked do you foresee sites in your environment will not be staffed?
A4: No. Rarely will you find a client that allows no staffing, operating lights out.
A3: At the end of the day, we all have customers. We’ll have to answer to them if we move away from the SLA they are accustomed to. So No.
A2: No. In our enterprise world, we do all add-move-change work as well as break-fix, so we’ll need 24x7 staffing on site.
A1: I don’t see us going to a fully un-staffed data center.
RVL: This matches what we are seeing as well. Clients are not anticipating going to fully lights out data centers.
A3: Past history traditionally has been everything on-prem. Now it’s a hybrid model we are reconciling and what we’ll look like in 5 years. When you put it in that lens, the public cloud touts the use of AI and machine learning as a selling point and option – cutting edge. However, it seems like an unproven science at this point. “If you want to reduce staffing levels, maybe going more to the public cloud helps this.” It’s a factor we’ll consider.
A2: Carbon footprint is another factor to moving to public cloud. Using AI for us is to improve efficiency and reduce carbon footprint. Cloud an option and potential way to do that. However, we do not want to be the tip of the spear.
A1: When I hear AI talked about by cloud providers, it appears most AI is being implemented outside of data center operations at this point.
A4: The NoVA Community College program is in its 3rd year. 41 students in the program so far. 20 students passed the program and have full time data center jobs. Idea of specialists in data centers has gone, so you need to move away immediately to generalists. Most data centers are looking at consolidating responsibilities. Overall, this is proven to reduce headcount by 1/3 across the portfolio. Jobs have migrated from trades to more administrative with more leadership and organization skills required.
A2: Asking now internally, do we start relying on vendors/contractors more and move away from self-performing, self-deliver staffing model? Our area is now struggling to bring in electricians. Also with 50% of staff planning to retire in 5 years, staffing with only specialists becomes a problem.
RVL: We’re seeing the migration to more generalists. We’re seeing technicians watch over vendors more and overseeing maintenance while trained to perform operations. However, in some areas vendors seem to be in short supply.
Using generalists, how do you deal with certain operational tasks that require electrical/mechanical skills?
A4: A generalist background doesn’t mean they are not qualified. They still need to be knowledgeable in the skills required to maintain and operate all facets of the data center.
A3: Internal training program is key to make sure they have all the required skills.
Anyone using analytics to drive maintenance activities and find problems before failures?
A2: We’re looking at it now. Goal is to drive efficiencies, failure avoidance. Will require adding a lot of sensors.
A1: Been using data analytics for several years and have moved to reducing and shaping maintenance around this. Cooling optimization programs are a product set that’s out there as well and seems to be effectively being used. We have not yet moved to AI where machine learning changes the set points.
Note - Other known data center operator education programs mentioned during the session:
• 7x24 Exchange, Carolinas Chapter - National Consortium for Mission Critical Operations at Cleveland Community College - see link below
• SMU – see link below
In this roundtable, the focus was on what members are presently doing in the New Normal, and what they plan on doing around data center staffing. Have you changed your staffing model to reflect the New Normal? Do you have a staffing plan and does it address staff availability through 2025 and beyond? Are you presently or planning on using automation and AI as a way to help offset staffing needs? What other staffing innovations are being considered? Rich Van Loo, VP Operations Services for Uptime Institute attended to add his experiences and insight.
Roundtable comments (presented in the form of a conversation with attendees):
Attendees were across 4 major verticals (financial, retail, telecommunications, and colocation) and provided their insights and comments on their present onsite staffing plans and the role automation might play.
Attendee 1 (A1): Presently operating fully staffed, 24x7. However, we are seeing a bit of staffing issues because of our close proximity to other data center sites in the area. Present concern over using automation and losing staff.
Attendee 2 (A2): Trying to figure out where we’ll be 5 years from now. Big part of evaluation is how automation will impact staffing. Also do we stick with the specialist model we presently deploy? Can we shift to more generalists with automation? We’re having issues finding specialists.
Attendee 3 (A3): Interested to hear about automation, AI, machine learning. Sense is we’re on the bleeding edge and not really seeing some of these things applied. Are others further down the line implementing this type of approach?
Attendee 4 (A4): Presently teaches introduction to data center operations at Northern Virginia Community College. Emphasis on entry level type of folks to fill staffing needs.
Rich Van Loo (RVL): Clients just starting to look at automation as a way to reduce onsite staffing, having people on-call, but not yet implementing for the most part. Sites are implementing more detailed monitoring at the sites as a part of this.
A2: We’re fully automated today with BAS/EPMS/SCADA/etc., with full monitoring that goes to onsite staffing. Presently down to single technician specialist during off hours. Our present goal is 2 technicians Monday-Friday, 24x5. All systems are remotely accessible now as well but requires onsite staffing to interact. We do want to get where we can allow remote interaction.
A3: We’re keenly aware of the shortage of qualified personnel. That of itself, however, is not driving staffing decisions. Confidence this issue of lack of staffing is being addressed and the pipeline will start filling. We are looking at if we have to have reduced staffing, how we deal with that during a sustained period to operate remotely – a lot of attention into that. We’re open to understanding how technology can help.
Question was then asked do you foresee sites in your environment will not be staffed?
A4: No. Rarely will you find a client that allows no staffing, operating lights out.
A3: At the end of the day, we all have customers. We’ll have to answer to them if we move away from the SLA they are accustomed to. So No.
A2: No. In our enterprise world, we do all add-move-change work as well as break-fix, so we’ll need 24x7 staffing on site.
A1: I don’t see us going to a fully un-staffed data center.
RVL: This matches what we are seeing as well. Clients are not anticipating going to fully lights out data centers.
A3: Past history traditionally has been everything on-prem. Now it’s a hybrid model we are reconciling and what we’ll look like in 5 years. When you put it in that lens, the public cloud touts the use of AI and machine learning as a selling point and option – cutting edge. However, it seems like an unproven science at this point. “If you want to reduce staffing levels, maybe going more to the public cloud helps this.” It’s a factor we’ll consider.
A2: Carbon footprint is another factor to moving to public cloud. Using AI for us is to improve efficiency and reduce carbon footprint. Cloud an option and potential way to do that. However, we do not want to be the tip of the spear.
A1: When I hear AI talked about by cloud providers, it appears most AI is being implemented outside of data center operations at this point.
A4: The NoVA Community College program is in its 3rd year. 41 students in the program so far. 20 students passed the program and have full time data center jobs. Idea of specialists in data centers has gone, so you need to move away immediately to generalists. Most data centers are looking at consolidating responsibilities. Overall, this is proven to reduce headcount by 1/3 across the portfolio. Jobs have migrated from trades to more administrative with more leadership and organization skills required.
A2: Asking now internally, do we start relying on vendors/contractors more and move away from self-performing, self-deliver staffing model? Our area is now struggling to bring in electricians. Also with 50% of staff planning to retire in 5 years, staffing with only specialists becomes a problem.
RVL: We’re seeing the migration to more generalists. We’re seeing technicians watch over vendors more and overseeing maintenance while trained to perform operations. However, in some areas vendors seem to be in short supply.
Using generalists, how do you deal with certain operational tasks that require electrical/mechanical skills?
A4: A generalist background doesn’t mean they are not qualified. They still need to be knowledgeable in the skills required to maintain and operate all facets of the data center.
A3: Internal training program is key to make sure they have all the required skills.
Anyone using analytics to drive maintenance activities and find problems before failures?
A2: We’re looking at it now. Goal is to drive efficiencies, failure avoidance. Will require adding a lot of sensors.
A1: Been using data analytics for several years and have moved to reducing and shaping maintenance around this. Cooling optimization programs are a product set that’s out there as well and seems to be effectively being used. We have not yet moved to AI where machine learning changes the set points.
Note - Other known data center operator education programs mentioned during the session:
• 7x24 Exchange, Carolinas Chapter - National Consortium for Mission Critical Operations at Cleveland Community College - see link below
• SMU – see link below
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