Keeping the lights on part 2: How do we optimize our energy usage?
John Frey (00:03):
Academics have surmised that if all we did in technology was switch our applications to one of those two programming languages, C and a programming language called Rust, we could cut technology power consumption by 50%, 50.
Michael Bird (00:19):
John Frey (00:20):
I don't know if I know enough to tell you whether that stat is valid, but even if it's in the right order of magnitude, that's a huge opportunity.
Michael Bird (00:30):
In the last episode of Technology Untangled, we looked at the ways in which we can generate new energy to meet our soaring demands as a planet. From liquefying biomass to wave generation and accepting the continuing role of fossil fuels for a few years to come, we took a long hard look at the ways we make energy, but how about the ways we use it?
Aubrey Lovell (00:54):
After all, using less means we don't have to create as much. Well, that's the focus of this episode, how to better use the energy and technology we already have today. We'll be looking at two sides of the coin. Firstly, the IT industry, it generates 6% of the world's emissions and affects pretty much every industry. After all, businesses need servers and frankly, it's the most on-topic area for this podcast. Then on the flip side, we'll be looking at our personal sphere and what's around the corner when it comes to being energy conscious at home.
Michael Bird (01:30):
Hopefully we'll all come out of this inspired to change our behaviors, reform the way we work, and with altered perceptions about the way that we use electricity, which is a pretty lofty ambition. So we'll also settle for being a fascinating journey into saving cash and the planet.
Aubrey Lovell (01:58):
You are listening to Technology Untangled, a show which looks at the rapid evolution of technology and unravels the way it's changing our world. We're your hosts, Aubrey Lovell.
Michael Bird (02:07):
And Michael Bird.
Aubrey Lovell (02:15):
Right now we need a lot more energy. To recap on a few stats from the last episode in 2023, on average we use 60 kilowatt hours of electrical energy each per day. That's enough to boil approximately 600 liters of water, which is a lot, especially when you consider that several billion of us are using a lot less than that. Demand is growing too by 3 to 5% globally every year. But making more energy, even renewable energy isn't a green proposition. A lot of raw materials go into making solar panels and wind turbines, so maybe we could use less or at least use it in a smarter way, but how? Well we talked about IT being one of the biggest users of energy. So let's start there.
John Frey (03:01):
I'm John Frey and I'm HPE's Chief Technologist for Sustainable Transformation, and that's a practice I actually started 22 years ago. And what I do is I work with customers to help them use their technology more efficiently and to share HPE's leadership. And that came out of, by the way, customers increasingly asking us that, "Your company makes all this technology, can you show us how to use it more efficiently?" And at first we handled that on a one-off episodic basis and then we realized, wait a minute, these customer inquiries are increasing and they're getting deeper and customers really don't understand how to drive efficiency.
But the other thing that was really interesting is in many cases they were turning off or negating many of the sustainable features that we were incorporating into our products. So part of it was even helping the customers understand the features that we put in their technology to make it more efficient and encourage them not to negate it. And even though I'm a technologist, I will tell you, the best answer of the problem is not automatically a technology solution. There might be some very low tech solutions that are just as effective.
Michael Bird (04:12):
And incidentally, if you're interested in technology for good and bridging the digital divide, we've done an episode of the podcast on using tech to end world poverty, which covers these issues. So do check that out.
Aubrey Lovell (04:30):
So what are some of these low tech solutions? After all, this is hardware which can cost hundreds of thousands or even millions of dollars. Surely running it more efficiently is going to be a task requiring a level of specific expertise. Well not always.
John Frey (04:45):
Look at the infrastructure and the technology that you have and make sure, if nothing else, make sure you're using the efficiency features built into it. So for example, if you think you've got a customer base or an employee base of 100,000 people, each of them have notebooks. Those notebooks all come from the manufacturer with power savings enabled. Now what happens in large enterprises for example, is they put their corporate image on the device which may or may not have the power management re-enabled. It's the same way for the enterprise gear and the data center. We work very hard on power proportionality, which is how does the device throttle its power consumption to match the workload that's running on it? But if the customer turns those energy saving features off, they negate many of the benefits that we've offered to them.
Another great example is one of the reasons why data centers often have much more cooling than they need, it's because IT executives know that when the systems get too hot, they start failing. HPE has been adding thermal sensors to critical components for years. One of the tools that we give to customers to get those devices on the network, HPE OneView will show them the data output from those thermal sensors and show them exactly what their thermal situation looks at. They can see power consumption. So they actually have the tools in many cases to resolve these challenges themselves because the manufacturers have put them in the technology, but so many customers are unaware that those features exist in the hardware they already own. So for me, that's where you start.
Michael Bird (06:26):
So manufacturers are in many cases already providing efficiency solutions for your data center, for example, and that is a great start.
Another company providing tools and technologies to encourage energy efficiency is VMware. So here's Joe Baguley whose job title is, well I guess we'll just let him spell it for you.
Joe Baguley (06:51):
I'm the Vice President and Chief Technology Officer for Europe, Middle East and Africa at VMware, which I think is probably one of the longest titles I've certainly had in my life. What we do at VMware is we virtualize servers at the core of our platform. And what that means is we enable someone to run 10 servers on one physical server. And so if you think back to when we first started over 20 years ago in 1999, and then really when it started to hit data centers in the early part of this century, the whole focus was, well, hang on a second, I've got 10 servers sat here all at about 10% utilization most of the time. What if I could find a way of running all those applications on one server? And that was really what we did with virtualization was we made one server look like multiple servers is the best way to think about it.
And initially people were doing that for cost saving, "I have to buy less servers." And then suddenly people realized that actually over the last 10, 15 years that there's an energy saving element to that too. So when you look at what a customer can do, there's a few things there.
One of them I'd really like to highlight first off is just basic consolidation of virtualization. But of course I'm going to say that work for Vmware. People are looking at some very simple stuff. If you're running two redundant data centers, well you're running a data center with another load of redundant stuff and another data center just in case with disaster recovery, but why not use the cloud for that? When you come to need it, it's there, you can spin it up. So I think the challenge is, to be fair, Michael, is that you've got to get people to think differently.
The real shift here is shifting operations models and shifting from we've got to have two of everything and a spare. I mean, how wasteful is that? So how can we in software do operations that mean we don't have stuff lying around idle just in case? And so if you think about how we can do that, takes me back to, we did a study a few years ago, this is before VMware again, my friends and I where we proved that with a couple of tier two data centers, you could get the same availability as a tier four data center for a lot less waste and a lot less money. And so this is the same principle, taking advantage of the technologies that are there today.
Aubrey Lovell (08:52):
Okay, so virtualized machines are a thing and a great way to save energy. They've also been around for about 20 years now. So we have to ask the question, what's the next big thing?
Michael Bird (09:02):
Good question, Aubrey, Joe?
Joe Baguley (09:04):
We've got a product called Aria Operations, which essentially is the operations tooling software that runs across someone's cloud operations and then what they're doing with their infrastructure with VMware. We have put into that this green score system and it's essentially a sustainability dashboard I think is the best way to look at that. And so the green score is really built up of different components. So if you're looking at how you operate more cleanly, you're looking at clean demand for a start, you're looking at a leaner operation and you're looking at a greener supply.
So how do I green up my supply? How do I get leaner in my operations? And how do I clean up the demand for that as well? So as we drill into those, there's sort of five areas that pop out. So cleaning demand actually comes down to sort of a workload efficiency and that really measures the data center's waste resources. And so that's looking at how much are we wasting in a data center of stuff we wouldn't need? So that's virtual machines that are idle and aren't being used by anyone, so why are they even powered on? Things that are too big. All of those we're looking at to make workloads as efficient as possible, so removing bloat in the workloads themselves, and we can do that from an operations perspective and that's exactly what we're doing.
Aubrey Lovell (10:14):
So I think this is really a fascinating conversation because we hear the term sustainability a lot in our industry and I think from the previous guests that we've heard from, we see that there is this alignment between understanding your infrastructure and how you streamline and can reduce and consolidate, almost aggregate if you will. How do we bring things on a smaller scale but actually have a better reward or more efficient workloads to be operated on without using as much infrastructure? To me, it's an interesting point. We think about IOT, I think that's a really big common theme here, Michael, where we are now adding to the equation, all of these devices. We talk about temperature sensors and all these tiny little devices that correlate into this bigger realm of infrastructure. Absolutely, it's important to figure out that model and how we can bring all this together.
Michael Bird (11:05):
Yeah. And I think people are more likely to make use of these tools if it's simplified, if it's all in one place. If you're going to have to open up a spreadsheet and read a sensor from here and go and phone up somebody else to get some more information, you're just not going to do it. Nobody has the time to do it. And having basically one location where you can get an idea of how efficient you are, how efficient your server's running, how the temperature of your data center, if you need to switch things off or spin things up, I mean that just is amazing. That is such a time saver.
Aubrey Lovell (11:41):
It is so nice because the outcome is basically when you can have everything in one dashboard or one place to monitor all of these devices, infrastructure, performance and be able to find those efficiencies, you then have time to go back into your business and focus on the things that matter for your business. You're not worried about your IT, you're not worried about your tech because you have this single pane of glass.
Michael Bird (12:02):
On a personal level, I'm certainly interested in how efficient my house is and how much energy I'm using. And even looking back 20 years here in the UK, I've gone from having no idea how much energy I'm using at any point in time to right now I can open up my phone and see exactly how much energy I'm using, how much it's going to cost, and that really enables me as a consumer to figure out can I be a bit more efficient? Should I switch things on a different time?
That's an interesting one because the way we as humans use electricity is a huge part of the puzzle both in the IT industry and in the world at large, right down to our individual domestic use. And that is an area where innovation and technology are also leading acquired revolution. Devrim Celal is CEO of KrakenFlex who are part of the Octopus Energy group.
Devrim Celal (12:57):
KrakenFlex is a software platform that manages all the relationship between physical assets and these can be electric vehicles, heat pumps, small batteries, generation assets like solar and wind, and ultimately utility scale batteries. And it helps utilities to connect, control, and optimize these assets against generation, typically renewable generation so that we're maximizing the flow of green electrons across the system, but also ensuring that we're using the cheapest possible electrons that we pass onto our end consumers.
About a decade ago we had abundance of what we call thermal power generation. So you burned fossil fuels and you generate electricity and those big power plants had one capability which was they could follow demand. So when you and I wanted to run our electric boilers or turn on the kettle, they would actually ramp up a little bit and generate electricity we need. Come to today, things have flipped upside down so we no longer have the control of when electricity is generated, so it's when the wind blows and sun shines, we have electricity. Therefore we have to get better at when we use electricity.
Now if I asked you, "Don't have a tea now." You would probably say, "Nope, I want to have my tea now. I'm going to have my tea now." And you won't change that pattern. But our relationship with electricity is changing as we get electric vehicles or as we switch from gas boilers to air source heat pumps. Suddenly we rely a lot more electricity. But the good thing about that is the pattern of that is shiftable.
So for example, immersed economies, people will arrive home roughly the same time, about 6:00 PM and people with electric vehicles when they get home will plug in their cars. That's absolutely the worst kind of behavior you want to encourage. Because it's when we're relying on fossil fuel generation, it's when the network is most constrained, and suddenly you're adding this massive load that's going to create a peak that we can't deal with. It's not about using less, it's timing when you use it. And because electric car is plugged in for 12 hours over night and the typical charging need is about 2 hours, so you've got 12 hours to schedule 2 hours of charging, why not do that when we have excess wind?
Michael Bird (15:26):
At like two in the morning or three in the morning when everyone's asleep?
Devrim Celal (15:29):
Absolutely. Especially if you don't have to get up and plug it in. If you have a trusted party like your retailer who will do that for you consistently.
Michael Bird (15:36):
It was quite interesting what he was saying about creating this relationship between the users and the suppliers, trying to figure out when users are using more energy, when suppliers have got excess energy. I don't know if you have this in the States Aubrey, but in the UK there's this whole concept of whenever there's a big football match or something big on the TV that happens, they have to, or at least they used to spin up power stations just to be ready for when the kettle got put on during the breaks. Just switching your kettle on multiplied over millions of houses across the UK meant that we had to switch on the whole power station just to be able to deal with the demand of it, which I think is quite fascinating.
That is a pretty analog, pretty simple way of doing it. Back in the times when we didn't have lots of data, we didn't have smart meters and things like that, whereas nowadays we have smart meters that are connected to the network that can communicate on how much energy is being used at any one time. And I think it's quite interesting that we can be a bit cleverer with when and how energy is being used.
Aubrey Lovell (16:41):
I totally agree. I think with the technology that we have today, it's so much easier to predict patterns and energy usage, especially around the grids, these smart grids that we have. And it's the opposite as well. So if you know that you're having extreme weather, think about today, we're breaking records for heat. So basically a lot of these smart grids, these power grids can actually be controlled where they start to reduce the amount of energy and efficiencies to protect the grid from any type of failure in terms of extreme weather. So you can see both elements. And I think one thing that's really key here, and I was kind of wondering this as I was listening, I was like, well, who would actually wake up to plug in their car at two and three in the morning? I think it would be up to the technology. You could create technology in order to automate it and do it for people. So how does it actually work and what was the motivation to connect every electrical device to some grand system beyond? How are we plugging it in essentially?
Michael Bird (17:42):
Well, it's all about resilience and creating systems to ensure a reliable intelligently used supply going forward into a future where renewable energy is a big part of the mix.
Devrim Celal (17:54):
All the technology we need is there, it's just a game of configurations. So on the physical side, we're hitting points where there are more electric vehicles sold per month than, let's call them traditional combustion engine cars or vehicles in general. We're increasingly getting more heat pumps being installed, especially the current environment of electricity pricing. So that's a check.
The next bit in that equation is the ability to connect to those assets. And when I say connect, connect them from a software platform to monitor and to control them. That part of it, we're figuring it out. That's the space that we're in.
The next piece of the puzzle is the markets. So when I look at wholesale power markets, the price starts at every half hour. That's been traditionally there. There's nothing to change in that, but there are these real time markets where National Grid, the UK system operator is making changes to allow smaller devices to participate in. So that's moving in the right direction as well.
The big part of the puzzle that we're starting to solve now, just a part of technology maybe that's not 100% there yet, but it's been in development, is the ability to move from a system where we used to manage a small number of big power plants to managing millions of smaller devices in the system. So the ability to orchestrate those, to schedule them at the right time. So we're helping balance the [inaudible 00:19:27] demand of electricity and that's where we probably go into more advanced algorithms like machine learning to be able to forecast when everyone is going to get home and plug their cars in. When they plug the car in, what will be the state of charge of that car? And what will each consumer want to achieve by next morning? They'll typically mostly say they 100% by 7:00 AM the next morning. So to be able to forecast when they'll plug it in and what will be the state of charge is one of the big challenges.
Aubrey Lovell (19:58):
So what we're seeing here increasingly is that firms at the cutting edge of energy efficiency are providing technical solutions to help us get the best results from our data centers or our devices, but that human buy-in is the key part of those solutions. And the tech is there to help us do better rather than hiding the issue out of sight and quietly fixing it for us. So what's driving this increased desire for change and to do better?
Michael Bird (20:23):
Well, you won't be surprised to hear that it's money or rather that it's because we've suddenly realized just how much running our technology actually costs us. Here's Joe.
Joe Baguley (20:34):
Way back in the day, even pre-cloud, the electricity bill for the data center was not ever seen by the IT department. So the IT department never thought about the consumption of the energy of the equipment. They would just turn up and say, "Here data center people, please rack this up and do it." And that energy bill was lost somewhere else in the company's operational costs. And so there was never this direct correlation. And so cloud really helped with that because cloud obviously then turned IT resources into a commodity, but when you start to commoditize something, cost becomes incredibly apparent because you have to cost it to do that.
So actually I think cloud's been fantastic in exposing cost models through to people and enabling to think a little bit more from an architectural perspective in terms of the full cost because when you run a cloud, you do cost the energy in. So I think that's really, really cool because you've got these fully costed models which are exposing energy costs. I was talking to a customer in the Middle East recently who was saying that they were having problems because their [inaudible 00:21:27] in Luxembourg was going up more than 2.5 times in price just due to energy. And so suddenly they were looking at us around consolidation, around how they can start to basically run what they're currently running in that [inaudible 00:21:37] on a third of the hardware.
Michael Bird (21:42):
John Frey has found similar results from his conversations. Frankly, at least in the IT space, the amount of electricity that's being used has always been, well somebody else's problem. And it's only when those costs start to get exposed that the argument to change things can be made.
John Frey (21:59):
What gets measured is something that gets managed. And so historically, if you look at how people are measured, IT leaders tend to be measured on their service level agreements, their uptime often, being able to deploy technology solutions for their company. That's what they get measured on. If you ask the question, "Well, who actually gets measured for power reduction?" It's real estate or facilities types of teams. So in technology, what we found was, and I think this is generally true, that the real estate teams, so for example, the folks that own the data center facility and the cooling and systems in that have spent a lot of time optimizing because their number one thing they're measured on is utility reduction or utility cost reduction. They've done a lot of work.
From a technology perspective of the technology products within, I've had CIOs tell me, "I don't see the power bill. Even if I wanted to, I couldn't manage it because I don't have any visibility to it." And I've had some say, "Nor would I because it's not my main KPI." So I can recall a very big insurance company that I went in to brief their technology team and I said, "If we're going to come in, let's invite your real estate folks, let's invite your finance folks, all those organizations that I mentioned a minute ago." And we actually had the briefing then in an auditorium and I was talking about power savings reduction and the technology executive said, "Well, I don't care anything about that. I'm not measured on that." And the real estate and finance leads almost simultaneously stood up and said, "We care an awful lot about that, so let's figure out for our company, how do we come together to manage all of these things together?"
Aubrey Lovell (23:45):
It's interesting. I mean obviously we talked about how everyone has their own goals and their own objectives, and I think for IT execs with the themes and the importance of sustainability coming into the picture for technology in the IT industry, this is something that they have to care about and think about more overall and performance wise, how they align to it.
Michael Bird (24:07):
I think the tech industry has traditionally been from a resource perspective, been pretty wasteful, but completely understandably because that's not their priority. Their priority wasn't originally to save energy or necessarily to save resources. It was to achieve a task. Didn't really matter where the electricity came from. Our knowledge and understanding and awareness of how important it is to save energy wasn't really something that was on anyone's agenda and it certainly wasn't on a board of a company's agenda.
Fast-forward to today, I think sustainability is a really important thing on most boards' agendas, and of course when we are thinking about our own infrastructure and how much energy we're using, how efficient it is, it's going to be one of the most, if not the most important thing to be thinking about.
Aubrey Lovell (24:58):
It is. And getting buy-in from diverse parties is an interesting factor because it goes well beyond what goes on in IT. The fact is we've got all of our own motivations and goals based on our situations.
Michael Bird (25:10):
Yeah, absolutely. And that's just as true to us individuals in the domestic market as it is to divisional heads in major organizations. You're just talking about much, much larger numbers of stakeholders, but how do you bring whole populations on board with the idea of saving energy and therefore money? Well, as Devrim explains, the key is both to provide people with the basic information they need to make intelligent informed decisions and perhaps more importantly, to make those decisions worth their while.
Devrim Celal (25:51):
Recently, there's a new national grid service called Demand Flexibility Service, DFS and Octopus Energy has been leading that with what we call saving sessions, where effectively we tell our consumers tomorrow at 5:30 or 6:00, the system is going to be tight, which means that supply and demand, available generation and what we are going to require is quite close. We'd like you to use less electricity, and we sort of create a game. We basically tell you, "Michael, we forecast at 5:00 PM you would've used 10 kilowatt hours of electricity. Here's the game, let's see how low you can go." And in the last two sessions, we've been able to get 200,000 people to participate and reduce their demand by 100 megawatts, and that's 40 massive wind turbines all generating at the same time on demand effectively. And it's brilliant because these people aren't making big money. They would've made a few pounds and they could either use that as a discount on their bill or put it into a fund that we created that we'll use towards consumers who are struggling to pay their bills in this difficult period.
So important part isn't the money they generated, it's more that they're so willing to participate to help the system balance supply and demand. And I think that's how the future of the energy system is going to happen.
So let me use this to give you a little bit of background, a framework to think about this. So if we think of energy tariff evolution in that analogy, our version one would've been fixed rate tariffs. So it was fixed, you knew how much you were going to pay regardless of when you use electricity. And then there was Economy 7, but that was about what you got. So with Agile Octopus, which was our first time of use tariff, Octopus Energy changed it a little bit and gave consumers a price signal that said, electricity really in the markets are priced 48 times a day every half hour, and here's how it looks and you decide when you want to use electricity. And that became quite a powerful tool to influence consumer behavior to the extent that Octopus at one point was shifting around 500 megawatt hours of consumption from peak to off-peak times, which was quite critical in the system.
Aubrey Lovell (28:09):
We talk about human psychology and behaviors and turning this into a game is quite an interesting approach because you're then giving incentive for people to be more conscious of their energy usage and trying to do better with consumption. And I know that you mentioned that you did this over the wintertime, so can you tell me a little bit about your experience with that?
Michael Bird (28:30):
Yeah, so my energy provider sent me a message and they said, "Look, tomorrow at six till seven or five till six," what you would've assumed would be the peak time of day for people to be using electricity here in the UK, switching ovens on or whatever it is. "If you use less energy than your average, then we'll give you some credits or tokens," whatever it is. I mean, it worked really well on us and from what I read, lots of people across the UK were doing this. So it seems like, I don't know why we haven't done this sooner. It sort of seems quite genius just to sort of gamify it a little bit.
Aubrey Lovell (29:01):
I'm really fascinated by the psychology behind that and if it helps take some money off of your bill, absolutely, I think it would be successful. So the key to moving on energy efficiency is human buy-in. Got it. But what can you actually do to save electricity and operate as greenly and efficiently as possible? Well, the first option is obviously to switch to renewable sources of power. It seems obvious, and in fact it's KrakenFlex's whole reason for existing, but it's not that simple when it comes to transitioning, say the IT industry. The bones of the solution are surprisingly similar to Kraken's offer, but on a mammoth scale. After all, it's a body that generates 6% of the world's CO2 and operates 24 hours a day, rain or shine. The solutions need to go further and beyond the realms of power management services as John explains.
John Frey (29:52):
We constantly get asked this question, as the world tries to shift to renewables, lower carbon energy sources, for example, what do you do when the sun's not shining and the wind's not blowing? If you wanted to power a data center using renewables, which is always the challenge, how can you do that knowing that the power generation varies? And so there's been all sorts of work on load balancing. So not every workload needs to run during the week, during the day. We know wind generation, for example, up in the evening, so perhaps if we can shift some workloads, you can do a better job to balance. But [inaudible 00:30:29], because they also do solar, wind research, fuel cell research and other types of renewables, they said, "Well wait a minute, what if we had renewable generation capability? And what if when we had more power than we needed for the technology stack, we bank that," they store it as hydrogen, so they actually break down water into hydrogen and oxygen, store the hydrogen, and then they power the data center using hydrogen.
They actually take two industrial truck motors, fuel cell motors, put them in an IT rack, brought hydrogen into that as a primary power source, and the fuel cells power IT racks on either side of it. And it turns out that that works really, really well. And then you don't have a tremendous amount of copper cabling coming in. And also because the hydrogen is a primary power source, the grid can be the secondary. So you don't need large uninterruptible power systems, football pitch size rooms full of car batteries as a backup source. You don't need the diesel generators outside as a backup source because you have a primary and a secondary. So it solves many of those issues.
And then the other thing was it turns out fuel cells get more efficient, the warmer they get. So if you're doing liquid cooling and you're coming off, your IT is warmer water, what if you wrap that warmer water around the fuel cell a couple of times to help it get even hotter, drives the efficiency of the fuel cell up, which warms the water even more. And it turns out there are some cooling technologies that you can take warm water and use it to create cool air for the last couple of things in the data center that probably aren't liquid cooled. So you can even generate cooling air from the warm water. And frankly, I don't think we've scratched the surface on some of the relationships that could have there.
Michael Bird (32:23):
Joe Baguley agrees. VMware as a company is carbon-neutral and puts a huge amount of energy, pun intended of course, into sourcing renewable energy for its operations worldwide.
Joe Baguley (32:35):
Being 100% renewable globally, it is very, very hard. It's not easy to do in the way that you think. You can't just suddenly go and find a solar farm next to your office and go, okay, fine, we're brilliant, we're sorted, we've got a solar panel. We will build one next to every office. So a lot of places that's exactly what we've done. So yeah, we do have, in Palo Alto, we have onsite solar. In fact we run a whole microgrid there to power that. And there's been some great advances we've done with that in terms of batteries and other bits and pieces properly running that as its own self-sustaining microgrid for our campus there. Say for also where we work with the local electricity utilities where we can to provide renewable guarantees or get renewable supply. So again, that varies in countries. So Bangalore, we've been able to do that. In the UK, it's quite easy to engage with a renewable energy supplier and buy your energy from a renewable energy supplier.
Now you and I know the way that grids work is that, yeah, I'm sure there's some non-renewable stuff going into our sites, but the point is, as long as you're paying, as long as 100% of the electricity you're paying for is coming from renewable sources, that shifts the grid and it shifts things in time. So for our US consumption, we purchase [inaudible 00:33:41] generated by US-based renewable energy power plants. So we try and change the country that you're in because I mean you've got to change the countries you operate in. And so your people always ask here, for example, at home I'm on Octopus, 100% renewable energy. And they go, "Well, but it's not renewable coming in your house." I said, "But no, but Octopus only buys renewable energy from the grid suppliers." So if we get more and more people only buying renewable energy, then that will just drive up the requirement for ... And so on, you've got to start somewhere.
Michael Bird (34:06):
Now, we promised this episode isn't an advert for Octopus, despite the fact we've got Kraken who are part of the Octopus Energy group in this episode, and Joe as a customer. That was just a very serendipitous link. And of course other suppliers are available.
We mentioned power management services a moment ago. And as it turns out, software could play a huge part in the solution in two ways. Firstly, because it can act as a control layer over huge swathes of energy usage as in the products offered by HPE, VMware's Aria System, and of course Kraken.
Devrim Celal (34:46):
We were all software people. We were new to the energy industry, but we could see the problem around the horizon, which was increased renewables would've required a demand site run system. And for that, we have to create flexibility. So we had this idea that we could build a high performing software platform that can connect to millions of small batteries, help orchestrate their consumption pattern to create this flexibility to allow the system to operate with higher renewable penetration. That was the dream. And there were two options. You either became vertically integrated, so you built assets that provided flexibility and you traded them yourself. Or you went completely horizontal and you built a software platform that helps others do what I just described. And it was an easy answer. We knew how to build software.
So our iteration three was developing the core of what's KrakenFlex today, a software platform that would be asset agnostic. So you can manage consumer owned electric vehicles and heat pumps as well as doing utility scale solar, wind and batteries, and optimize these at scale in real time and sell that as a software service to people who participated in this general marketplace, which would be infrastructure funds who developed and owned these, traders who traded them, retailers who provide the flexibility. Today with Octopus Energy, we're finally rolling what we call the true digital energy revolution where we've got tariffs in the market like Intelligent Octopus where consumers can sign up using a mobile app, they register their electric vehicle, heat pump, battery using the app and give Octopus Energy the ability to monitor and control those.
Aubrey Lovell (36:35):
So good software can save you energy and save you money. But on the other hand, software can have its challenges if not designed correctly. To be more precise, there's a huge issue with badly written inefficient code. And if you might be thinking, well, sure, but how much economic impact can badly written code have? Well, here's John with a statistic that may surprise you.
John Frey (36:57):
Even the programming languages that we use can have a very material difference. In fact, there's a great academic study that compares 13 programming languages in terms of power efficiency, compiled time efficiency, runtime efficiency. So they're looking at a variety of attributes and what they found was the most efficient across the board. And they sort of trade places, number one, number two, depending on which metric you're looking at, are C and a programming language called Rust, R-U-S-T. And if we think about then when we're writing code, we use a more efficient language. And although it's not my statistic, I'm just going to report it, but the academics have surmised that if all we did in technology was switch our applications to one of those two programming languages, that we could cut technology power consumption by 50%, 50.
Michael Bird (37:52):
John Frey (37:53):
I don't know if I know enough to tell you whether that stat is valid, but even if it's in the right order of magnitude, that's a huge opportunity. And when you couple that with the fact that we know that about 70% of business applications today are not enabled to run in the cloud, and yet so many companies want to ... Whether or not, by the way, and I don't mean public cloud in this instance, I mean just a cloud type environment. So we know that many of these business applications are being rewritten so that they are able to run in a cloud environment. So if you're going to rewrite them anyhow, our advice is pick one of these efficient languages. So it's one thing to use the right language.
Then there's another aspect, it's called software performance engineering, which is, and again, remember I'm not a programmer, but programmers tell me that often when they're coding, if they have a routine that they need to write to do something, they'll first look at a software library and say, "Has anybody written code already to do that exact thing? If not has anybody written code that does something similar?" They'll just take that bit of code from the software library and then modify it as need be to do what it is they want to do.
Now, that's from a time perspective, much more efficient. They didn't have to write that code from scratch, but perhaps that might be 10, 20, 30, 40 lines of code to do something had they written it to do just exactly what they wanted, might've been 2 or 3 or 4 lines of code. In fact, I had a coder tell me earlier this week at an event I was speaking at that she was able, when she did this software performance engineering on code she was writing, she was actually able to get her code to a 10th of its size, original size, so reduced down by 90% just by going through and doing some software performance engineering on it.
Aubrey Lovell (39:46):
Wow, a 50% reduction, 90% more efficient coding. That's pretty amazing. But it also shows that there's always room for improvement.
Michael Bird (39:55):
Yeah. And that's no reflection on developers or coders. As John said earlier in the episode, you're not historically being judged on how efficient your code is because well, you aren't paying the bills. But if those numbers are anything like the average, that could soon change as Joe explains.
Joe Baguley (40:16):
The developer never sees the impact, the direct impact of any decision they make about how they code things in terms of energy wastage. They might see something runs a bit slower, but that's different. They'll just go and use more resources to make it run faster rather than looking at running more elegantly. And so actually, when we were doing the Code of Conduct, one of the things a bunch of us talked about in the British Computer Society was there will be somewhere in the future where developers would look at developing energy efficient code. Now, the interesting thing is why would a developer want to do that? Well, they do that because you start exposing it through cost as you talked about. So those economic factors start to come into play. And so then developers started to realize that, "Yeah, it costs me money to run my thing. It costs me running to spin instances up, et cetera, et cetera. So I need to work out how I architect to be as efficient as possible based around the cost."
Aubrey Lovell (41:04):
So there's two solutions. Use better energy sources and use better code. And then of course, we've already talked about simply using less technology and turning things off when you simply don't need them.
Michael Bird (41:15):
So I guess the final question is can tech ever really be sustainable? Here's John Frey to see us out.
John Frey (41:22):
I think the technology industry can be sustainable. In fact, the Global E Sustainability Initiative, or GESI as they're called, did a report called Smart Cities 2020 and Smarter Cities 2030. And basically their conclusion was that technology has a great ability to decarbonize. And in fact, their statistic says it's about a 10 X decarbonization. So for the power and the embodied carbon that you have in the technology solution itself, you can get a 10 X return decarbonizing other places. So I think we can be. Part of the challenge, and actually, I was on a panel conversation in New York earlier this week, and one of the questions that came from the audience was, "That'd be great if when technology got more efficient, we stayed at the lower consumption level." They said, "The reality is every time we find a way to decarbonize or lower the footprint of technology, we just find more things to do with technology. So aren't we tugging at both ends of the rope, as it were?"
And so I think we have that opportunity as well. We've got to ask ourselves, do we really need all the technology that we take advantage of? If we know that when a European footballer posts a picture on Instagram and it gets liked and commented on millions and millions of times, and the energy budget that takes is equivalent to a small city, is that really a great use from a societal or environmental perspective? I'm not sure how we solve that challenge other than to continue to drive efficiencies. But I think technology is a great enabler of carbon reduction. If we look across all the carbon commitments that companies have made around the world, the experts tell us that perhaps we have solutions to about 50% of the commitments that have already been made. So as a technologist, and as an innovator, I get excited when I think, wow, how are we going to solve some of these other challenges in ways that either don't exist today or we haven't thought of today yet because we haven't brought the right minds to bear?
And this is for me how this concept of bringing diverse points of view into the mix really pays off. Because if you put a bunch of technologists in a room, we're going to find a technology solution. But if you bring artists and inventors and entrepreneurs together, there may be some solutions that derive from that we would've never done as just a group of technologists, for example. And that's what I love about that diversity of thought. There's a term called catalytic collaboration, which is how do you bring this variety of voices together and sometimes voices that haven't been heard historically, and can you get better outcomes for bringing that diverse groups of voices together?
Michael Bird (44:13):
And that is something I think we can all agree on, that the solution to how we use our energy will be one that takes a huge amount of expertise, and each and every one of us. You've been listening to Technology Untangled. We've been your hosts, Michael Bird and Aubrey Lovell, and a huge thanks to our guests, John Frey, Joe Baguley, and Devrim Celal. You can find more information on today's episode in the show notes. This is the fourth episode in the fourth series of Technology Untangled. And next time we're exploring the amazing world of exascale computing. Do subscribe on your podcast app of choice so you don't miss out, and so you can check out in the last three series.
Aubrey Lovell (45:00):
Today's episode was written and produced by Sam [inaudible 00:45:03], Michael Bird, and myself, Aubrey Lovell. Sound design and editing was by Alex Bennett, with production support from Harry Morton, Alicia Kempson, Alison Paisley, Camilla Patel, Alyssa Mitri, and Alex Podmore. Technology Entangled is a Lower Street Production for Hewlett Packard Enterprise.