Greg Irwin  0:18

Good to speak with you all. Thanks for joining us. This is our latest session of the SETC, the sustainable enterprise technology council. We're partnered with WWT. And the idea behind this forum. This series is really, you know, driving practical discussions, steps, actions and community around, you know, practical sustainability. There are some pretty, I'd say, large scale impressive goals for carbon neutrality. It's it's a massive undertaking. And what we're thinking about here is, how do you actually start to chop it up into problems that can be solved. And we run these monthly discussions, we are publishing research, and we're building community. So we're trying to do all of that in the form of zoom. And at least right now, and, and keep and keep the conversation growing, I'm going to pose one task for everybody, it's a very simple task is to make a connection across this course. It's, it sounds kind of simple, but please follow through on it. Look across the grid, it doesn't have to be WWT. But I want to encourage you all to take advantage of others, and the experience of others who have common interests. So go through LinkedIn, or ask us here and we'll be happy to help connect people. But please, the your goal out of this session, if nothing else, is to make one new contact. In terms of logistics, we're going to use the chat window throughout, please. You know, the, the more active the chat window, typically I think of it as the more engaged you all are, we can have a successful session with you sitting back and just listening quietly. But I have a hunch, this group is pretty smart, pretty outspoken, and pretty, pretty proactive. Let's make sure that we're getting dialogues going on the sidebars. So I'll encourage you, What's one way that you're thinking about using data analytics around sustainability. And then the SETC, I'll put the overall ask somebody here on the WWT side dropped the URL for our website, make sure you sign up for the newsletter. And that way we can keep this. Keep this conversation and pace going. With regular dialogue and regular plugs. So please, I'm gonna ask you to join basically, you know, follow what we're doing in SETC. And of course, pass the word word along if you have somebody at your organization, or a friend or colleague who you know, cares a lot about sustainability. We'd love to continue to build down the group here. So any referrals are highly appreciated, greatly appreciate it. Let's get into the core topic here. And let me let me bring in Brendan Walsh. Brendan, do us a favor, give a little bit bit So Brendan and I are are shoulder to shoulder here in terms of driving SETC. Brendan, share with us a couple comments in terms of where you see SETC going.

Brendan Walsh  3:48

Thanks, Greg. So really, really excited about tonight. So this afternoon's topic is obviously very contemporary, in a DC datacenter analytics, and specifically energy consumption within the data center. And obviously, not just the consumption, but the emissions that come from there really are at the heart of every sustainability discussion, every ESG plan, and certainly every Net Zero plan. And so this topic is particularly relevant and timely. You know, I was thinking about this URL there. And I think sometimes its simplicity, it saves the day. And what I mean by that is what you'll hear from David is, you know, ideas that are straightforward and hopefully simple to grasp. And even more is a not 100 to implement. But just to touch on a point that you made about the magnitude of the task when Feeling was here. To put things in perspective, if all the data centers in the world were our country, they have been ranked number five by emissions. So that's pretty significant. Second point to bring this to life is the emissions from data centers are now very, very close to those from the airline industry in total in total. So when you think of the number of the 1000s, the hundreds of 1000s of flights per day, and the emissions that come from that the emissions that come from data centers are coming up very, very close. So I think that gives us some idea of the magnitude of what we're dealing with. But to get back to reality, David is going to outline for us, you know, some of the the key analytics that he has used in the past. And hopefully that will help bring things to life and stir a lot of discussion. So without further ado, let me turn it over to to David.

David Locke  6:08

Thank you, Brendan. Yes, it can all hear me. Yeah. Good evening. And good afternoon, everyone, wherever you may be. I appreciate the the introduction there in terms of sort of what we're covering. And certainly, whether this is, you know, the data analytics part is you need data, you need the initial data to even perform, you know, whatever calculations you need. And we're going to come into sort of where, where that information is coming from what we're seeing in the market, I want to keep this fairly sort of interactive, but also just just sort of informal in a way of like the topics that we're seeing across our customer base. And I think, as Brendan said, a lot of this is just keeping it simple and looking at where you can start in these places. So I am going to share some material in this particular session. And I'll do a sort of more longer intro in terms of my own background. So I'll just pull this up to start with. So we were talking about this being about data analytics, I would say it's more about data center metrics and data and information that we are well hearing and seeing, certainly our customers asking for, or looking to measure, and even understand just, you know, where are they today. So that's really what we're covering today, says my background. So interestingly, I actually did start my career in a utility company. Those in the UK would know of eon, or if you're in Germany, and I was an architect there as they moved into the energy trading business back in the early 2000s Way from the original Paul System. I then moved into consulting and tenure with Accenture, which also helped develop EMCS consulting practice all around the data center and data itself. And then I moved into that organization sort of full time. And then most recently into web t, where I'm a Chief Technology Advisor, covering all kinds of topics across infrastructure in the data center, and cloud, in terms of operating models, etc. And working with very top tier financial services organizations, in terms of their strategy for it, but also most recently, around the whole topic of ESG for it, so less so on, say, financial ESD and corporate ESG. But this is around, what practical steps are we seeing and hearing within the data center? So we're going to start with what are we actually talking about here? And what is the issue, I guess? And what is the challenge that we were trying to solve? What is the problem and the exam question. And ultimately, it comes down to the products that sit within a data center. And I've got two examples here, one from Dell, and one from HP. And both organizations recently, and some of you may have come across this in other sectors and other markets as well. But we're seeing a trend now of manufacturers of which Dell and HP are clearly very large ones in the IT industry actually providing information to the market about their product, carbon footprint, sort of terminology. There are some other names being used for this in terms of the sort of lifecycle calculations used for sort of, you know, a particular product, but essentially, they all are using similar methodologies. And it's about the manufacturing of the device itself. It's about the transportation of getting it to where it needs to get to, and then its its usage. And that's really, what you'll start to see on this chart is where the biggest impact is. And so while these are separate studies of very similar technologies, you can see that the the output is actually very similar. So if we break this down, you can see you know, on the Dell slide it's bit easier to read on the pie chart, but they put roughly 15% of the overall emissions for one of their servers is to do with the manufacturing process, and actually creating that product. You know, within the factories, the transportation of actually then getting it to a customer or to a reseller, or distributor is tiny, you know, in terms of the actual impact required to move that particular platform, and they're not particularly heavy devices anyway, you'd know. But it's the usage, that really, is the lion's share of that products. impact in a lifecycle. Now, I should say, the studies in both these cases were done over a four year time period. So it's estimating the overall usage of electricity over a four year period. Where it's a fossil fuel based electricity supply, you know, otherwise, that would actually be zero, if you're using renewables, but we'll come on to why that's not always beautiful. Now, there's some examples there in terms of what the equivalent footprint of one server is, versus driving a car, or planting trees, or even our own carbon footprint as, as individuals and humans, you know, no impact on the planet. And the numbers start to sort of add up and get quite scary. So this is the kind of like baseline we're working with. This is the issue, effectively, the data centers and our customers have. And as we said, you know, it's a, it's becoming a bigger portion of not just the global consumption of electricity, but also, certainly within an organization. It's a significant portion. And we've seen it anywhere between 20-25% of the total consumption all the way up to 40%. In some, some examples, depending on you know, how did it or what how IT centric, those companies aren't. So that's kind of sort of a starting point on this topic. And what we're really trying to do is work out how we can reduce the the emissions that that data center is, is pushing. And some of the techniques we can apply, to bring ourselves to net zero. So let's look at this chart in more detail. Our goal from for everyone is to get to net zero. And that's defined by each organization to their science based technology. Metrics, as I'm sure you and your organization's are doing, too. And you have an overall footprint that that device and that environment is contributing. And there's a number of things we can do to help this. Now, most organizations, some organizations that I say, could just pay to get rid of those emissions, and says the whole talk about offsets about carbon credits, or even more altruistic options, such as planting trees to offset that activity. Regardless of what that offset is, I think the message I'm going to be sharing and this is, that's the last thing you want to be doing, you know, you want to minimize the amount of permissions you have in the first place before you start looking to to offset and we'll look at that impact. You can use renewable energy. And so we talked, you know, that that usage, aspect of a particular product is, you know, 70-80% of its overall emissions over a four year period. If you can plug into a renewable power source, whether that's when you're supplying yourself on site, or you're getting that from the utility companies as an off site, contract, then you can zero out your consumption and emissions from that demand. But again, that's not always possible. And again, we'll sort of look at some of why that is. Now it's the first two that we're going to focus on, which is really around how do people use the technology? How can they be better sort of citizens, corporate citizens, it citizens tend to how they use that equipment? But also what can you do? Or can you do things as sort of quick wins just to improve the environment, that these things run in, as in, you know, some things to do with the building? And how it's how it operates in this? So we talked about renewables. Why is renewables? Not the answer today, but anyone you know, that is aware of the different generation types in terms of fuel. This is from the US up to April of this year, renewables is increasing. Absolutely. And it is continuing to grow globally. But it's only going to be it's only 20%. Today, of all electricity generation in the US. And it's similar in different countries and first of all, environment. And there's still a heavy heavy reliance on gas, as we all know from the issues with gas supply and gas prices at the moment, as well as coal with a bit of a backstop. Iran nuclear. And that's probably something for a different conversation. But we'll see how that helps balance this out in the future. But certainly the gas and coal is a problem. And with that, if there's only 20% of the electricity available for renewables, customers, but your data centers are using 20 to 30% of your consumption and a customer, you know, you can't even get all the renewable energy you would need just for your own organization. So that's the gap, we've got to try and bridge until we can get more renewable energy in the market. And going back to offsets, you know, so we talked about there being the data center, if you added all the data centers up together, it was kind of I think, Brendan, you said it was like the fifth largest consumer or sixth largest consumer of electricity globally. According to Google. This week, there's roughly 100 million servers deployed. And if we take those metrics that we saw on Dells, product, carbon footprint slide, you can see that this is, if you multiply this up to the 100 million platforms, this is an inordinate amount of comparison, you're talking 1.2 billion acres worth of trees, so enough trees to cover half of the entire US, we're talking about the same emissions as half the population in their lives, in terms of that environment, or it would be the equivalent of taking every car off the road in the US for a year to actually offset. And I think, Brendan, you've said it, it's like the equivalent of the airline industry, in terms of the suit, you know, these are the comparisons. So we've really got to, this is why this is such a big issue, and why we need to tackle it. So we start looking at these behavioral examples. And, you know, everything comes down to what you use, how you use it, how are you measuring it, because you can't fix anything, or you can't change anything, if you don't actually measure that in the first place. So what we're seeing and what we're advising, and what we're looking to share and get your thoughts and insights on as well, is really, what are the datasets we have? What are the datasets we need? And how can we make the right calculations to actually create some insights into this space? So across the IT industry, there's a number of different tools and different things you can apply to the estate, we can look at utilization, we can look at the efficiency of a particular server, how much power it's storing, but workloads is it also running? And for what purposes? You know, does it need to be running 24? By seven? Or does it need to just need to cover the business hours of the day? How are you actually acquiring technology, you know, we know many customers and there's we're moving more and more towards consumption based contracts or operate, you know, OPEX based commercials, because certainly, traditionally, a lot of customers would take their, their budget annual budget. And they will be buying technology upfront, and then they will be growing into it. So you imagine you buy 100 things, and you plug them all in, and they're all happily plugged in operating, but you're only actually using one to 3% of that environment as you actually start to add your users into it. It's not an efficient use of that technology. And so moving more to the OPEX model and these consumption based contracts can is a way to solve that. Alternatively, you use the you know, the cloud. And so we all know that you can leverage AWS, GCP, zero, etc. And in the most part, they've all committed to fully green IT services. And so they're already they're already delivered by renewable energy, or they already have plans in place that allow them to offset correctly, those emissions as part of their, their operating environment. But there's other things we can do and have people you know, from a behavioral point of view, like understand where their workloads live, how they operate, can they be right sized? Are we actually maximizing the amount of our purchases, and about how we did it and I think, looking at the statistics, the average utilization of a server inside an enterprise organization is around 30 to 40%. Whereas actually, if you use a cloud provider, and there's an example there in terms of their sort of storage services, you you're really paying for what you you know, and they will manage the overheads and the typical efficiency level increases up to like 85 90% of that environment. And so, these are all these are all best practices and targets to to aspire to. And so certainly within the the average enterprise organization, there is a lot of opportunity to consolidate, update, modernize, and generally take the Pour actors out to the environment. And I think we've probably all got examples where there is something sat in a corner somewhere, it's plugged in, we don't really know what it does. But no one really wants to switch it off just in case that affects the, you know, the business in some some way. But that's no good anymore. We've really got to do proper assessments, proper audits, and truly understand, you know, what everything does, and is there a different way of doing it, whether you're rewriting applications, you're moving to the cloud, you're using someone else's application, you know, a SaaS service like Salesforce, or workday, and all these other sort of newer platforms that have emerged in recent times. So this is the sort of environment we are sort of living in. And I've got a couple of examples here and modernisation in itself. And it might be quite low level, it might be quite sort of even specific it but it shows an example of how technology is actually is helping to actually help itself, you know, in this example. So this was an example from a large banking customer used a lot of EMC storage that was very dense, very heavy, used spinning, hard drives, lots of power consumption, lots of floor space. And they've recently changed, suppliers change vendors, to a more modern platform that happened to be flash based. So using your SSD type storage, so no spinning, mechanical operation, just everything on solid state infrastructure. Now, it was a bigger state, roughly 80 of these big machines. But the comparison, moving from the old to the new, was roughly a 10x reduction in electricity in floor space in the carbon that was then emitted from that. So there's things we can do to look at what you already have, let's see what else is available in the market? Can we consolidate? Can we upgrade, you know, can we move to something more efficient? And these are the sort of numbers that came out of that assessment and that study, and these are not small numbers, as I'm sure you can imagine. So there's a cost saving itself, but then that translates equally into a carbon reduction in terms of that electricity consumption being reduced, overall. And even if you were using Yes, please do.

Greg Irwin  22:20

David I’m sorry. I'm gonna jump in with a question if that's okay. And I know there are a number of other questions in the chat coming here as well. I'd like to stay on this for a moment, because I think there's the presumption based on this analysis that modernization will drive efficiency will drive energy efficiency. I'm curious if this client actually saw a reduction in electricity usage, and, and carbon output or if they viewed this modernization as an opportunity for savings on their expansion. So it ultimately was still a net increase in electricity and carbon carbon outlets?

David Locke  23:04

Yeah, it's a good question. So I mean, like for like, there is a clear reduction, just not just in writing, but just in physically, it has much smaller footprint, it doesn't consume as much as many power outlets, etc. So there is a clear reduction. However, they didn't make that choice based on that, as you say, it was more just on the commercials and the financial benefits of moving to a different platform. And actually, they're still growing in all other facets of their environments. So the datacenter is still going up. And actually on that point, we've got one of our other customers in a similar position where they're looking for, they would like to reduce total datacenter consumption by 4%. year on year from now until 2040, which is their net zero time. The reality is and data center managers spoken to I said, if he can hit zero, he would be very, very happy. And because they've always been growing, you know, so how do you reduce something that is already growing at the same time. So this is why it's more of a sum of the parts type game, we need to be removing the things that you don't need, you can use other people's data centers colos. And you know, there's an example here because you pay coalos based on consumption, so that electricity consumption, and while it might be renewable, and therefore there's a low, low emission value related to it, there's still a high spend, you know, because that's not how they're charging, but and then covering they've got to cover their renewable costs at the back end. So I think it's a very good question in terms of like the overall and the the other thing with these very large customers and I don't know if anyone else is it would be in a similar position, but a lot of them are buying their electricity in advance. So they're on big annual prepaid, you know, they're buying by the Giga watt hour contracts gigawatts not terawatt. So these very big contracts. And they probably have thresholds in terms of what they do and don't spend over the over that time period. So this is just think I just picked up, I'm seeing the chat there as well. So yeah, in terms of the SSD, hazardous waste, so you're absolutely right in terms of that manufacturing element, there may be a difference in terms of the actual embodied carbon that lives within that new product, because it's using the solid state technology as opposed to spinning this technology. But again, that until the market catches up, and all the vendors are providing these PCF type documents in a common format in a common standard. We're still sort of having to map and match things manually at this stage. And this, this is really for me, part of the reason for this this forum, I think it's excellent. Is that need for standards, and actual common metrics and methodologies that can be shared, is kind of one of the key things to this being a successful thing across the industry. I think everyone's just waking up to it. And I've got an example of that. Coming up soon. Do you make any others? Great,

Greg Irwin  26:21

let's keep at it, let's judge, on the one hand, use this as our mile marker to keep going through the conversation. But if others have questions, please feel free to use the reactions. And you can raise a hand, I'd like to try and keep us on track. So we can make sure we hit the key points around data analytics, otherwise, certainly this topic, we can go in lots of different directions. But I want to encourage questions. So either in the chat or raise a hand, David, you've, you've got the floor.

David Locke  26:50

Okay, no problem. So I'll move on, because I've got too much more. So this is exactly what I was just saying in terms of actually a positive change in the industry. So this was actually announced this morning, over at VMworld or BM EVO, what whatever they've decided to rename it over in California this morning. So the new version of VMware and if people don't know what VMware is, it's what runs the virtualization technology, in data centers to run virtual machines so small. So it's being able to use single service to run multiple workloads simultaneously. And it's been a very successful platform, I would say, 99.9% of customers, if you've got VMware in their environment, as it's a way of consolidating and maximizing efficiency of infrastructure in the first place. They've actually added in this new version, disability to monitor the energy consumption per host, whether it's active or non active, so you'll now be able to see, why have I got something switched on, if it's not doing anything, you know, I'm actually going to actively track when workloads Come on, when they come off, what parents using when I'm idle, what power I'm using when I'm at peak. Now this starts to become interesting when you start looking at it from a data analytics point of view. And appreciate, you know, this topic, I'm not going to go into a deep dive on sort of the the true analytic aspect of this, but because there are platforms, and there are tools being developed tools to start using, these metrics being being being built, as we speak. But these are the kinds of metrics that we're looking to capture from environments, so you can make those informed decisions. We want to be able to know like everything about how our systems are running, when are they being used? Where are the hotspots in our data center, know, where have we got certain areas that are cooler than others? Could we actually move workloads and one of the benefits of VMware and tools like it is you can actually move the application from one physical server to another physical server while it's online. And actually just use that as a way to consolidate and actually get to a point where you could be switching whole racks of equipment off overnight on an automation script. And then bringing it back online once it's required, you know, in the morning, once the users come back online, so this is very much for me that data. This is the deep literally day one of the kind of data analytics for this sort of market and being able to track Power Systems consuming energy in the data center, and we've been obviously looking at some of the areas we want to use. So to sort of summarize this, if we go back to my chart here in terms of the glide path of getting to net zero, it all starts for me kind of like knowing what you have. You've got to look at the baseline. You've got to look at your operational sorts of activities. How you may obtaining that estate and we've spent years and years, creating environments that are very secure, very highly available, very resilient, which means often you're buying more than you need. And you're running dual sites, you're running dual clusters and all these kinds of overheads. But we really need to get into a different model now, where you're designing and architecting for sustainability. And not necessarily for reliability, although that is clearly still a critical element. But it's like, how do we do that? Do we have, you know, a primary site within our organization, and then use the cloud as the second site? Or do we just find better ways of architecting or consuming applications from other vendors, you know, from from SaaS provided, where you can just pull this stuff down over the internet. And then it's already a hosted service rather than having to do your own it. The plant and equipment upgrades. So we haven't touched on this like in detail, but this is all getting into, you know, better cooling of the environment. It's hot, and cold aisle containment, were able, it's bringing in liquid cooling, or even immersion cooling, I noticed something around crypto scrolled past earlier, but the ability to actually compare cooling technologies, it's something we're investing in here at WWT, you know, the difference between air cooling versus liquid cooling versus immersion cooling, depending on the type of workload and the type of technology you're using are going to be critical. Moving forward, we probably will see a whole datacenters emerged in liquid in the future where they're running very high processing, AI and ML type workloads, or crypto, mining, etc. And so this is, again, a huge area of being investigated, as well as all of the automation to actually change the profile of the data center or the equivalent depending on the requirements and meet. So that this is a huge area of innovation at the moment. And we're and I think there's some people on this call from some of the organizations we're working with in this space. But it's absolutely you know, and it's a lot of it, again, is about data, about having more things like smart power supplies, better analytics, in terms of pulling information off that mechanical equipment, putting metrics off the air conditioning equipment. So you can actually correlate between heat, power consumption, usage, temperature, etc. These were mashing together sort of two worlds that didn't really talk to each other previously, so it utilization and actual building and data, the building management utilization. And then we talked about, you know, some of the, the renewables, and what the challenges there that isn't enough, currently, and even if you were to add your own power generation to your buildings, it might be okay for an office to run off solar, but you're not going to run a data center of solar today, not with the efficiencies in that type of technology. So, again, it's it's, it's all going to evolve. And you know, and this is kind of the start point of how we offset that. And with the goal, then that we just, you know, we try not to get too upset as little as possible in terms of the financial end of this, this chart. So this, hopefully, it was just a sort of good overview and a good thought topic to sort of open this up, really the floor and look at sort of how others might look at this in their own environments. Anyone that's already gone down some practical steps to look at this and the quick wins that they've had you've had, and really anything that we might be missing in terms of how we get this. Now there's, underneath all of this, there's a huge activity around data platforms, and the ability to track information and the metrics coming both from the IT side and the the real estate side of the house, in terms of your user population where they are working from home working in offices, all this information. And so as the true topic of data analytics, absolutely, there is an engine and a source and a sort of a system of record, almost that's required. So in my mind, I often see this whole sort of carbon accounting, nag analogous with financial accounting. So you need to manage it in the same way that you have these like a general ledger and you've got your p&l. From a financial point of view, you've got to have a similar p&l for your emissions. What are you burning? What are you getting back? What are you recycling? And what's that sort of circular economy look like, over time? And so that needs to be, you know, stored somewhere it needs to be annulled. lie somewhere. And that's, you know, things that we're seeing in the market things we're doing ourselves as an organization to actually help capture, analyze, and then report and provide insights in this space. So that's sort of the formal part. But yeah, really just open it up for the floor.

Greg Irwin  35:18

There's a hand and we probably have about 20 good questions sitting in our sitting in our, our message board, I'm gonna go to my, my, my co host here, Brendan and and see if there's one thread that you want to start with his start a start us off here. Otherwise, we'll go to 10 do we'll go to Manny, and a number of the really good questions. And please keep the questions coming here. I do want to start bringing the group in for the dialogue. And I'm going to start with Brendan Walsh, Brendan any key questions you have for David?

Brendan Walsh  35:55

Yeah. So look, I've been following the chat. I think there are so many good questions, and so much good dialogue, I would just hit it right from the start and see if we can work through as many of the questions as possible.

Greg Irwin  36:09

Okay. All right. Very good. Let's, so there are a number here. Boy a pick, pick one. I'm gonna ask you to I'm going to get a task. Right. And what's one topic you want to you want to start with in the chat?

Brendan Walsh  36:27

They have we come across any innovative AI ml use cases in the data center optimization has been, you know, the the first one, what is the lifecycle?

David Locke  36:47

Read it out info overload.

Brendan Walsh  36:50

Well, actually, David, why don't we just do this one? Let's do this one, because it's very, also very pertinent right now. Another metric we see our customers are tracking is a waste management, waste by unit produced recycled to the whole circular economy. Yeah, we've talked about it before. Take that one. That's a good.

David Locke  37:11

We did. Yeah. So it's, I guess it's not really covered in what I was speaking about previously. But yeah, absolutely. Products have that three pronged part to their lifecycle. They've got their manufacturing, they've got their usage, and then they've got their retirement. And so we didn't really touch on retirement, but it's definitely a huge topic and becoming more and more relevant. And we've seen an uptick in the type of it disposal companies that will now do sort of full recovery of the REM metals, and elements, they're actually contained in most circuit boards, and, you know, most servers and other technology equipment. So that with the right organization, and the ability to recover, in some cases, as much as 100% of the materials and not put anything into landfill, they're also able to report back on what that effective avoidance is for reminding those elements out of the ground. And in fact, I saw an example report only today, where it'd been done for a financial services customer a review on, you know, a year's worth of technology disposal, and it came back with something like 245,000 metric tons of avoided co2, a further 100,000x of the water required to actually manufacture or so produce those materials. And there was another metric as well, I think around, we had water a hydrogen, as well. So we see are seeing, you know, other metrics is not just about the carbon footprint, it's definitely around the materials, it's about those raw, you know, and rare elements that exist gold, platinum, lithium, you name it, you know, these are not cheap materials, either. Certainly, when you have a bulk, you wouldn't throw away gold jewelry. So why do we throw away gold in service, you know, it's the same principle. So we're seeing that sort of messaging come through now. And as I say, organizations that are much more effective at recovering all of those elements, and not just sort of separating into your classic sort of metals, plastics, etc. At the base level, or retreat switches, just like any other piece of it equipment, they contain chips, and they've got a you know, power consumption. I think, what it what differs in what you sort of raised is you kind of need it anyway. So it's going to be something that might be on an operating regardless of what's traveling through it, you know, in terms of that connectivity, because you can't really turn a network off if it's under utilized. So I appreciate that is probably an interesting area to see how you might look at your connectivity and the different methods for different hours of the day. And maybe there is different ways of routing cost different networks based on those efficiency metrics. So again, it's a bit like we don't know what we don't know, until we find out, you know, and this whole ability to actually understand the baseline in the data center and what everything is actually consuming. What we're seeing at the moment is, it's kind of at the rack level. So most customers are either getting a smart PDU reading, or they're doing their own readings of the overall rack. And a rack could obviously have up to say, 30-40 components, and it can do electricity, to actually get the individual power consumption per device, and they know what to do with it. There's an argument between the effort and the value of what you're going to do with that. But I think the more interesting thing would be that behavioral thing of, okay, so I've got multiple networks, I've got multiple pods that I can use to get to my servers, you know, everything these days is load balanced. And as we said before, designed for high availability and resilience. So what if we could actually lower that sort of resilience level or the UN actually increased the utilization by switching off certain parts at certain times of the day? So I think it's more these will be the decision that you can make once you have that information, but it is a good question in terms of how that correlates. With the increases in speed that we'll see over the coming years to

Greg Irwin  41:21

We’re talking about the the overall carbon emissions, what have you seen in terms of the storage tier, either different vendors get techniques in driving efficiency?

David Locke  41:37

Yes, in terms of pure consumption, that big shift from spinning disks to flash has had a 10x reduction. So it's kind of done one exponential shift, but that the data still grow, as you say, and then density of SSD is getting higher, and it's still taking up less footprint, but it probably is consuming more energy to keep those things alive and running on a day to day basis. I think that if I go to the destruction part, though, I mean, some of the wasting and tying it back to the waste. And the destruction thing I get, especially working with financial services, you know, every disk has to be shredded and destroyed. But these newer waste disposal organizations are are able to still recover the materials, and they will drive specialized trucks to the data centers do the shredding on site, but then take the shredded material, you know, for the recovery periods, I think that part will be covered, you know, over time, as more and more organizations use that type of technology. But the growth in storage, agreed, but it is also becoming much more everything is running on a server somewhere, you know, and even you say about, you know, the difference between having a storage array and distributed storage. You know, there's one possible example is you know, everything becomes commodity, you're running everything on the same type of compute platform. And you're actually using all the pockets of storage for your capacity as much as you're using the compute for your your workloads. And I know that's not exact science. But this is sort of where things are obviously going. And so you do have more ability to distribute that load on to different service tiers, agreed take could welke make a comeback, for your kind of long term archive based on its relative minor impact, but it's an interesting one to take away. I think it's been tackled.

Greg Irwin  43:30

I want to come back to analytics for for a moment here. And and the core, the core topic, and I guess, David and Brendan and I posed to the group raise a hand Has anyone had real success by initiating an analytics program, pointing and, you know, tracking core metrics, highlighting an inefficiency. And as a result, you know, being able to, you know, kick off an improvement project. I would love I'm sure that's, that's the goal here. That's what we're talking about. Let's measure it. Let's find the disconnects, or the you know, the outliers. And then let's hit him. I'll start with David. But I would love to hear anyone else who has as a story they can share around the use of analytics.

David Locke  44:28

I do have a quick one, actually. And it ties back to one of the questions in the chat around Does this include cooling and no heat, heat generation and things that come out from the actual AC and HVAC sort of environment? So, quick story, we had a customer that actually had their AC units were running at 100%. And they were about to make an investment in more AC so they could put more equipment in the data center. So they asked us to do an assessment analyze that. We looked at the actual thermal I'm probably the thermal view of how the air was getting from the the HVAC across the servers. And what it was doing in terms of its ability to actually cool the servers while they were operating. So there's a lot of data collected included, like heat cameras, thermal imaging of the environment to understand where the air was going, what transpired is that the AC unit, were actually running too hot. So they were actually too powerful. And the air was going past the cabinets too quickly, and it wasn't cooling the server's as efficiently as it could. And so the net effect was the data actually said, No, the outcome was that the data said, if you turn your AC down, your IT equipment will be cooler, and you can grow without buying any more AC. That's pretty cool. So it didn't even require additional spend, it was actually a case of turning something down. But until they had the data, they didn't know that that was the right answer.

Greg Irwin  46:03

Please remember the one task, make a new connection across this group. And you can go directly or again, come through through us here reply, respond to those reminders, and we'll make the connections. Thank you all for the engagement in the chat. But let's keep these going. And, you know, the dialogues don't just have to happen here during during our sessions. A big thanks to David Locke. Great overview, great content. And, and a reminder for everybody. You want to connect with David, you want to connect with Brendan, they're doing this for awareness around what they're really pushing around sustainability. And of course, we're happy to connect with the team there. David and Brendan, any closing comments for the group?

Brendan Walsh  46:51

No. Critic discussion, Greg, thanks for organizing. Thanks for the content. David. I think hands are follow up here in the chat. We should endeavor to get back to people on thanks everyone. Bye bye. Thank you guys.