Subscribe
Share
Share
Embed
HPE news. Tech insights. World-class innovations. We take you straight to the source — interviewing tech's foremost thought leaders and change-makers that are propelling businesses and industries forward.
MICHAEL BIRD
I want you to take a moment to think about your computer. When was the last time that you shut it down? Like properly selected the shutdown option?
SAM JARRELL
Oh gosh. Um, I mean, outside of when you're doing updates to it uh, probably this past July when I went to visit family to go do my gender reveal for my baby and I was like, "I'm not taking my computer with me. I'm shutting this thing down and locking it up in my closet."
MICHAEL BIRD
Good shout. That's great. Yeah
SAM JARRELL
What about you?
MICHAEL BIRD
I'm a fairly regular shutter downer, um, 'cause I know that if you shut your computer down, often it'll run a bit better when you start it back up again.
SAM JARRELL
I guess if we like spin this idea up to the size of an organization, that's a lot of potential saving power, especially if you think about like data centers and things like that.
MICHAEL BIRD
E-exactly, Sam. And, and this week we're looking at how organizations shutting down their tech when it's not in use can save them millions, and why despite these potential savings, people are still afraid to do it.
I’m Michael Bird
SAM JARRELL
I'm Sam Jarrell
MICHAEL BIRD
And welcome to Technology Now from HPE.
MICHAEL BIRD
So we have spent decades building IT, focusing on maximum uptime, but with rising energy costs and increasing power becoming an increasingly scarce resource, efficient use of potentially limited uptime is becoming more and more important
SAM JARRELL
Okay, so we're moving to a world where the quality of uptime is becoming just as important as the quantity of uptime and this choice, I would imagine could lead to quite significant savings for organizations
MICHAEL BIRD
Yeah, exactly. and if an organization can use their infrastructure more efficiently and then turn them off when they aren't needed, the power savings can really add up over time, which is why this week we are talking to Jostein Birkeland,
Principal Technologist for Sustainable Transformation in the HPE Living Progress team.
Now, to sort of set the scene to begin with, he started by explaining how we've ended up in a world where people are somewhat hesitant to power down their technology. Is that right?
JOSTEIN BIRKELAND
I've studied this the past year, and I came to the conclusion of that there's three primary concerns. Uh, one is when you switch things off, the temperature, decreases and you have more temperature fluctuations, so you can get hardware failures when it goes back up.
two is not starting correctly on an application, level, so it's booting up incorrectly, which causes business, continuity impacts. And the third one is that you get, a lot of chaos, in the networking space. So in data center networking, you will have alerts of switching ports, going up and down, which is causing a burden for network managers
MICHAEL BIRD
the first one is, really interesting. I didn't know it was because of temperature fluctuations. I just thought it was the, hardware hates me.
JOSTEIN BIRKELAND
Exactly. Yeah, that's right. so that's on the hardware side
MICHAEL BIRD
I mean, those are three things that is a real deterrent for organizations to just be like, "It's just easier if we just leave it on from customers when you have these conversations.
JOSTEIN BIRKELAND
Yes. and also I think a reason that we're leaving these things on, if it's working, don't introduce an unnecessary risk. but also we're seeing that the people that know how to implement this and would, know how to set up processes to do this, they aren't incentivized to do so.
so IT teams, don't have a goal, or a team objective, to save power, to switch things off. so those who are measured on that in the facility side, they don't have roles that allow them to do this.
So it's not really the ones knowing how to do it that are measured on it or, asked to do so.
MICHAEL BIRD
Yeah. that's been my experience when I've worked with IT teams, they don't think twice about power usage or any of that sort of stuff. It's, it's really just about, uptime
So is there a shift there then from an incentive perspective?
are we seeing facilities and IT departments having to start to collaborate a bit?
JOSTEIN BIRKELAND
we're expecting this, due to, power prices increasing, and also the concerns around power availability. we're also seeing that when it comes to, queues, for, requesting power, in countries, data centers, have been put at the bottom of the list.
So it's gonna be more and more important to effectivize your IT, and to show that you're implementing measures to run it as efficiently as possible.
MICHAEL BIRD
it feels pretty radical 'cause what it sounds like you're saying to organizations is ignore all of those things you've been taught about powering things down and actually now consider powering things down. what sort of things do organizations have to consider if they're wanting to start powering stuff down
JOSTEIN BIRKELAND
Yep. So first of all, they need to consider which applications and what kind of workloads are in scope. and that really reduces the risk, in the cases we've worked on, it's, development, workloads and other non-critical, workloads. So if they don't, boot back up again correctly, there's a slight delay
And then what we've seen is very useful, is to have, scheduling frameworks set up so that you know when you have expected, tasks and workloads to be done and when you can actually identify periods, of idling infrastructure.
and then also of course assign policies to the different, hardware, pieces supporting those, workloads so that you have policies for what can be switched off and what should not be switched off.
MICHAEL BIRD
So, you talked about not necessarily switching off critical infrastructure. how do you make sure that critical infrastructure doesn't accidentally get turned off?
JOSTEIN BIRKELAND
By rules and policies. so you simply exclude it from, the workloads and the supporting hardware infrastructure in scope. So you only switch down workloads which have assigned the policy that it can be switched off. And thenin a virtual environmentwhen that,resource need then drops, then you start powering down the equipment, supporting that.
So you simply just have policies in place, on workloads which should not be switched off to not switch them off.
MICHAEL BIRD
Hmm. That's interesting. so are we talking about physically switching off hardware or are we talking about powering down, for want of a better word, like virtual machines, but the actual underlying hardware is still switched on?
JOSTEIN BIRKELAND
Good question. it's both. so it begins obviously on the software layer, on the virtual layer, switching off, workloads. another example is, VDI, for virtual desktops. So as you switch off your, VDI user or you power down a development, workload, of course then, automatically the hardware is not, switched off as well.
but once that resource demand goes down, then you can start powering down the physical, hardware as well. So, it's a two-stage approach, which includes both
MICHAEL BIRD
Got it. Got it. and, i-is it more valuable to be switching things off when they aren't in use, or i-i-is it better to be finding more tasks to keep the technology up and running all the time rather than it just sitting idle?
JOSTEIN BIRKELAND
Definitely more valuable to switch it off, when it's not in use, I would say. So the ideal scenario is really that when equipment is in use, that you ensure a high utilization rate of it, and that you are able to identify those periods of idling, or periods of very low expected utilization because very few tasks are scheduled for those moments, and perhaps be able to reschedule those tasks outside of that to ensure an idling period and then, switch it off.
but I would also like to mention, an entirely different, strategy, which is when you spot, periods of idling infrastructure that you can lease it out to other users outside of your organization, that needs these resources, on demand, or for intermittent, workloads, so CPU and GPU resources, for example.
So that's another completely different, tactic to ensure high utilization of the equipment, that it's, that's being used while it's on, and that it's not idling.
MICHAEL BIRD
okay. Can we talk numbers then? So, so when we're talking about switching things off, you-- we talked about there needs to be a good incentive.
what sort of savings are we talking about? how much power could be saved?
JOSTEIN BIRKELAND
So interestingly enough, in the two tangible cases I've been involved in, the number is 25%, so a quarter, of the power consumption of all the equipment, in scope. so to give a little bit more, context for the two cases, one is, internally in HPE, in our storage engi-engineering department, where they, have extensive scheduling frameworks, and start experimenting with switching things off, when there were no scheduled, tasks.
And the other one is with, one of our customers. it's a large financial institution, and they also had 25% power savings after implementing t his. and in that case, we had a total of 3,200 server nodes in scope, and during the weekends, 1,200 of them, were switched off.
The result of that is a 600kW power demand reduction – which leads to approximately one million euros saved each three years.
MICHAEL BIRD
And that's potentially relatively easy to save, but maybe organizations don't really think about it as an opportunity to save money there because the done thing, as we've said, like the done thing is always just to leave things up 'cause you don't know if it's gonna be coming back up again
So I guess this is now maybe turning away from it being a technology discussion, and maybe moving more towards the sort of conversation you'd have with the CFO. are you seeing that shift in who within an organization is starting to care about these things?
JOSTEIN BIRKELAND
Yeah. Traditionally in our team, we've mostly spoken to CTO, or data center managers, server managers. and in the last, years, more and more we're seeing new roles pop up. people in the IT organizations, responsible for sustainability in IT, for example. But, regardless of that, we've always tried to invite both, a CFO, a CTO, and perhaps a head of sustainability to the consultations we give to customers.
it requires a collaboration from the technical team, and then we can articulate the environmental benefits of that to the sustainability, lead and also the financial benefit of that with the CFO.
MICHAEL BIRD
What sort of difference could it make if this money was then invested somewhere else within an organization?
JOSTEIN BIRKELAND
Quite a lot of, impact and difference. so that money could, for example, be used to drive, modernization or renewal of old equipment. 'cause typically we see that companies are not able to do this because of lacking budget, when in fact doing that can, can free up even more budget. so if you imagine that you can start replacing your aging infrastructure, and consolidating onto fewer devices, you can have fewer support contracts.
You can also have fewer, software licenses. costs start, dropping. And I think this is one of the bigger, opportunities that companies have, in addition to switching off equipment, modernizing old equipment. one of the bigger opportunities they have to save, costs. And a good example of that is one of our global customers,
They changed their IT strategy in 2024 from a cost efficiency-focused strategy to a power efficiency-focused strategy, and started doing exactly that, switching out old equipment, with newer, more efficient, equipment. So they improved the power consumption per core.
so reduced that. and also got the total number of servers down
they had over 16 gigawatt-hours, of power savings in a year, which also translates to millions, of euros.
MICHAEL BIRD
that's really interesting
and
are there examples where actually workloads can be shifted to be done at different times, and like that's another way that could be used to save power?
JOSTEIN BIRKELAND
Yes, absolutely. Payroll is a good example. Um, it's a one that's one- once per month, but you also have, weekly tasks, for example, running backup jobs, or even daily. so scheduling these tasks, within the work week, that's exactly what we did, with this large financial institution that we started, planning for these backup jobs to run, Friday, before the weekend.
And then once that was done, powering it down, during the weekend days and then back up again after the weekend on Monday morning.
we're starting to see emerging interest in, load balancing, which is what you're speaking about, for peak shaving.
and if you think about it, it can sound a little bit like conflicting interests between ensuring high utilization when it's on and switching it off during, idling periods. That could contribute to, increasing the peak. but when you look more into it, you'll see that, when you implement scheduling frameworks, you'll probably be able to avoid the peaks, and still identify periods of idling.
MICHAEL BIRD
I mean, all of these sort of put together, it feels like there's loads of ways for organizations to save energy and, therefore save money. It feels like a change in mindset is required for organizations. is that something that we are starting to see?
JOSTEIN BIRKELAND
Yes. and especially due to constraints, in power availability and rising power costs, and also constraints in the supply chain. we see the customers are trying to understand how they can more effectively use what they already have, and how they can implement that efficiency also into procurement decisions
MICHAEL BIRD
do you think it's going to be a bit of a snowball once some organizations do it and then other organizations go, "Oh, okay, actually it's not that bad"? Like, do you think,this is gonna be the standard way that we as organizations approach power management?
JOSTEIN BIRKELAND
And that sounds lovely. I really hope so. That's also what I'm working to try to achieve. And I think the most important thing that we can do there is to, show examples of success, and also really address the concerns, around shutting data center equipment off, when it's not in use.
MICHAEL BIRD
I suppose we've sort of talked quite a lot about the theory but actually I think in practice this feels like it could be quite complicated, require quite a lot of scheduling and orchestration and making sure everything's working.
this stuff has to be rock solid, doesn't it? So how do you actually go about doing it?
JOSTEIN BIRKELAND
Yeah. So we're managing the private cloud of, of one of our customers that we implemented this with. and what we're doing is that then we're shutting down, workloads on Fridays. so every workload without an exception is then switched off. and with this customer, we actually had some challenges, in the beginning, triggering the reboot, on Monday morning.
But bear in mind that these are non-critical workloads. so what they had then were a few complaints, from different employees. what we did then was switching back on, equipment related to those complaints and leaving the rest off. so what that resulted in was that then 20% of the workloads that were shut off initially were never actually ever turned back on because no one were using them.
So it's also a good way to identify, unused applications, and things really idling for long periods of time.
how it also works, um, when things are booted back up is that,the equipment is also then running a health check. and if, there is a failure in that health check, then that server profile is then sent to another piece of hardware equipment which successfully, went through that health check.
So this results then also in a practice of preventive maintenance. So for example, memory modules then that fail in that health check can then be switched or replaced outside of active use.
MICHAEL BIRD
And I suppose the scheduled shutdown is something we’ve beendoing with wireless access points for years. So this feels like maybe and extension of that?
JOSTEIN BIRKELAND
Yes, and those Wi-Fi access points, they consume a lot less power, right? So it's less impactful, but there are many of them, and they automatically detect, if there is traffic, and if there isn't, it goes into a low power mode. And from networking equipment, and data center switches, we also have, features where traffic is monitored and then if there's less traffic, we can then shut down modules of that switch, so the line card supporting, groups of ports and switching that off.
So we do actually have, features in our tools for networking equipment to do this. So why not start, experimenting with servers and storage equipment, which, represents the vast majority of data center power consumption to do the same?
MICHAEL BIRD
Amazing.
Jostein, thank you so much for joining us on Technology Now. It's been an absolutely fascinating conversation
JOSTEIN BIRKELAND
Thanks, Michael, for inviting me
SAM JARRELL
Okay, so one thing that really stood out to me from this conversation is that this actually seems like less of a technology problem and more like an organizational one.
MICHAEL BIRD
Yeah, I think the same thing. Like maybe almost an industry-wide problem, 'cause I think I was always taught you never switch things off. because you never know if it's gonna come back up.
I remember somebody talking about how spinning disk hard drives, like they just don't like to be switched off. where possible, you don't switch them off and switch them back on, switch them off, switch them back on, 'cause that will cause hardware failures.
when Jostein was talking about this, I was like, "But, but you're not supposed to."
SAM JARRELL
It like creates some natural sort of like uncomfortability, right?
MICHAEL BIRD
Yeah, because, it also goes down the if it ain't broke, don't fix it. if it ain't broken, don't switch it off. just leave it. Just leave it be. And I know there are so many IT departments where everything is held together with string and sellotape and hopes and prayers.
but we're in 2026, infrastructure is changing. we need our infrastructure to do more. and as Jostein said, there is, power, like power prices are going up worldwide, and power availability is also, in some instances restricted.
SAM JARRELL
I think, you hit the nail on the head with talking about "Oh, if I turn this off, it may never come back on." But in your conversation, with Jostein, it seems like sometimes the things don't actually need to be turned back on, right? he said something like 20% of the workloads for one of the groups he was working with, they did never come back on because it turned out no one was using them.
And if you think about that on like a mass scale, that's so bad.
MICHAEL BIRD
that is incredibly common.
there are so many workloads, so many devices that I'm sure have been idling for many, many years that just aren't being used. So there's just incredible opportunity here, but it requires, as you said, a mindset shift. It requires a different way of thinking
SAM JARRELL
Yeah, I think it, it requires incentivizing different things.
because right now I think, yeah, the wrong incentives are in place. It's keep everything up and running for as long as possible just in case or because we're a little bit scared versus, okay, let's see what happens because the potential savings for energy costs, everything, are far greater than the actual risk of just a little bit of downtime.
MICHAEL BIRD
you can see it from an IT department's perspective. They don't wanna lose their jobs, and if stuff goes down and never comes back up, like you would definitely worry about losing your job, wouldn't you? So like the path of least resistance is to not rock the boat.
But if IT and facilities are becoming closer together then actually if the incentives change, then the mindset changes
SAM JARRELL
Yeah, I agree. And that's exactly why I keep coming back to, I think this is an organizational problem because as you guys were talking about, this becomes also a conversation with the CFO
MICHAEL BIRD
I think it's a really, really fascinating new development I mean, we've done episodes on these, huge mindset shifts that we as our organizations are having to make when it comes to the way we think about our technology, and this is no different.
this is a huge mindset shift with the way that we're gonna have to think about what our technology does, how it runs. and I think it's a good thing. I think it will mean that we're using our technology significantly more efficiently
SAM JARRELL
Oh yeah, it'll mean tons of savings and I with the rising costs of everything, and particularly power, I think that this will eventually become standard practice. It's just gonna go through a really awkward, uncomfortable period where
some places are going to make mistakes and there's going to be some growing pains through all of it. But then eventually it'll be like, okay, no, the best practice is we turn stuff off during this scheduled period, and here's the backup plan for things, and if we have to, here's what we do to make sure that everything is recovered properly.
but for the next couple of years I can foresee this being like, okay, we're gonna take the plunge and turn off the data center and see how people handle it
MICHAEL BIRD
do you remember we did that episode, and the person we interviewed mentioned that, AI has changed our, expectation on how quick we get answers to things or how quick technology moves.
Because with, chatbots, like you don't get the answer straight away And so the concept is actually, we can slow things down a little bit.
And so maybe in parallel to that, you put a request and the server boots up and actually maybe we have this mindset shift of okay, if it takes five minutes for it to boot up, I'll wait five minutes to boot up, and then it will do the, the thing I want it to do.
Anyway, in my interview with Jostein, there were a lot of numbers and examples in it. So what I wanted to do was finish with a bit of a recap. A sort of if there's one thing you take away from this episode, this should be it. So the final thing that I asked Jostein was to just summarize the biggest incorrect assumptions which exist today when it comes to powering down our tech.
MICHAEL BIRD
Now my interview with Jostein had a lot of numbers and examples in it, so I wanted to finish with a bit of a recap – a sort of “if there’s one thing you take away from this episode, this should be it” so the final thing I asked Jostein was to just summarise the biggest incorrect assumptions which exist today when it comes to powering down our tech.
JOSTEIN BIRKELAND
It would be the assumption that the benefits aren't greater than the risks when it comes to switching off data center equipment. if you just do it for non-business critical workloads, the risks are low. and as we saw in that one example, you can then also identify things which aren't even in use and save, that very unnecessary power that goes to, powering that hardware
SAM JARRELL
Okay that brings us to the end of Technology Now for this week.
Thank you to our guest, Jostein Birkeland
And of course, to our listeners.
Thank you so much for joining us.
MICHAEL BIRD
If you’ve enjoyed this episode, please do let us know – rate and review us wherever you listen to episodes and if you want to get in contact with us, send us an email to technology now AT hpe.com and don’t forget to subscribe so you can listen first every week.
Technology Now is hosted by Sam Jarrell and myself, Michael Bird
This episode was produced by Harry Lampert and Eva Higginbotham with production support from Nik Damarell, Beckie Bird, Finlay Macrae, Alyssa Mitry, and Jenessa Ayache. Our theme music was composed by Greg Hooper.
SAM JARRELL
Our social editorial team is Rebecca Wissinger, Judy-Anne Goldman and Jacqueline Green and our social media designers are Alejandra Garcia, and Ambar Maldonado.
MICHAEL BIRD
Technology Now is a Fresh Air Production for Hewlett Packard Enterprise.
(and) we’ll see you next week. Cheers!
SAM JARRELL
Bye y’all