Episode #59: Going Green with Serverless with Paul Johnston (Part 1)

July 27, 2020 • 53 minutes

In part 1 of this two-part episode, Jeremy chats with Paul Johnston about how serverless compares to traditional computing in terms of being "green", the impact of data centers on climate, why efficiency is only a first step, what people in tech can do to affect change, and so much more.

Watch this episode on YouTube:

About Paul Johnston

Paul Johnston is an interim CTO, CTO and strategist who has particular interests in serverless, cloud, startups and climate change. Formerly, Paul served as a Senior Developer Advocate at AWS for Serverless and CTO of multiple startups, including one of the world’s first serverless startups. Paul is also a co-founder of ServerlessDays. Watch this episode on YouTube: https://youtu.be/SI2-WU_0zgs


Jeremy: Hi everyone. I'm Jeremy Daly and this is Serverless Chats. Today I'm speaking with Paul Johnston. Hey Paul, thanks for joining me.

Paul: Thank you very much for having me.

Jeremy: So you are a consultant through Roundabout Labs and a research associate at the Leading Edge Forum. So why don't you tell the listeners a bit about your background and what you have been up to lately?

Paul: So, yeah, background's a bit confusing. It's always a little bit strange. I don't have this whole 14 years at any one big company or anything like that. I spent many years in tech, 20 years, working with various different startups from my own business, that kind of thing. Then I worked in a startup in 2015 that effectively started using AWS Lambda.

So this is where the serverless comes in. And I was one of the first companies to start using Lambda in any kind of scale in a startup as a kind of first principle. And then I went from there to using it in that startup in 20 countries. Went a bit mad, a tiny, tiny budget from AWS. And I was like, "Well, this kind of worked so I'm going to keep doing it," and started telling people. AWS took notice, gave me a job, that was quite fun. Was a senior developer advocate for serverless at AWS for a while.

And then didn't stay there all that long, but it was really enjoyable while I was there. And moved away from there to go and do some consulting, which I've done since 2018. 2018? 2018. And then from there...

Jeremy: What year is it again?

Paul: Honestly, this year has gone on for a very long time.

Jeremy: Right, right.

Paul: And since then I have done some consulting in various different projects, tech projects. But one of the things I've done is worked on working out how tech and climate work and how they intersect. And one of the projects I've been working on is a research project for the Leading Edge Forum, which if any of you know Simon Wardley, that's the organization that he works for.

And I've been working on a project to look at how climate change is going to affect business over the next 10 years from a tech angle very much, so from a data and a tech angle. And just trying to see what lessons we can learn and what things are going to be coming up in the future. So kind of many and varied, shall we say?

Jeremy: All right. Well, listen, I have been wanting to get you on the show for a very long time, because I think this whole climate change thing is hugely important. I have two young daughters. I think about their future. I think about the junk we pour onto the earth, the pollution, the amount of carbon dioxide we're creating.

And one of the things that I think really attracted me to serverless in the beginning was not just, you know, obviously not having to manage servers, which is great. But this idea that maybe by sharing tenancy on a big server and only using the compute that we needed to, I was thinking in the back of my head, I'm like, "Well, maybe that reduces the amount of energy we use."

And so I know you have dug into this tremendously, and I mean, you're an expert on this stuff. And so I'd love to go through all these things with you, just get your insights, get some thoughts on this stuff. We can talk about serverless and some details of serverless as well. I'm sure that's what the listeners want to hear, but I love this idea of going green with serverless. Because I think it's hugely important. I think it's a step in the right direction.

But maybe we could start and just, or start by saying, how does serverless technology compare to traditional technology or traditional servers when it comes to green computing?

Paul: So it's a very, very good question. It's almost impossible to answer in some ways, but it's really, really easy to answer in others. So one of the things you want to look at, first place you want to start is, well, effectively you want a definition of what serverless computing is.

Paul: So let's just kind of take function as a service is kind of the base enabling technology, shall we say, for most serverless computing. Because I think most people will kind of see serverless and they'll go, "Right. What does that mean?" And so you want to drop it down to something that is kind of tangible.

So you want to talk about function as a service really as being the base enabler, because serverless for me is about business value and getting as much as possible out of your technology in terms of applications and all of those elements.

And so I think when you talk about function as a service, what you get is you get a pay for what you use. So you know that you are using as little electricity for your application as you possibly can. And so what you're trying to do is go, "Well, I want to be as green as possible." So being as green as possible means actually reducing as much of your usage as possible.

That's essentially what we mean by being green is actually reducing and actually using as little, as close to zero, in terms of compute as we possibly can. So what does that mean? How do you do that? Well, in terms of building an application, don't build the application to start. Just don't build the application at all if you possibly can. If you can build it on the basis of lots and lots of caching or not running any servers at all, then great, do that.

If you can do it on the basis of only running compute when you absolutely have to, then great. If your application can scale down to zero and it literally can, nothing is running if nobody's using it, that again is... That's the kind of thinking that goes into being green and being serverless, which is why serverless is something that for me works really well alongside an environmental conversation. Because it's not just about what does the techno- how does the technology work?

It's actually, well, this approach allows me to say, well, it gives me business value. It gives me environmental value. And actually when you come down to it, it just works out as better common... It's more common sense when you actually try. And when you build the application, you come out to the other end, it's usually a better application and easier to build going forward as well. So you've got all of these things working positively. It just seems to... It seems to work out better.

Jeremy: Right, right. Well, I love that idea of deciding whether you even need to build the application. That's a really good way to just cut your carbon footprint is say, "Don't even build the application."

That's super easy. But unfortunately there are applications that still need to be built and we still need power. I'm sitting here with my 16 inch MacBook Pro drawing 100 watts of power with the fan spinning a million miles an hour because my CPU is going nuts and just generating a ton of heat.

Obviously in the data centers, you have millions of computers that are generating a lot of heat, that are drawing a lot of power. And like you said, every time there's some execution that needs to be done, that CPU has to spin up, which means we need more power, caching, SSD drives, things like that - very low, low, I guess, wattage types of equipment like that obviously draw less and there's ways for efficiencies there. But there's only so much of that we can control. We're still going to need to do compute.

So what about data centers themselves? What type of impact do those have? Because I think it's more than most people think.

Paul: Yeah. So data centers are, let's put it this way. They're a huge impact. And they're a significantly greater impact than most people realize. And I wrote a white paper in 2018 with a friend of mine, Anne Curry, and she and I did an awful lot of digging.

Now when we did the digging, we found out an awful lot around people saying that it was going to be, there was going to be a tsunami of data, which meant that data centers were going to grow at a huge rate. And we found that it was going to be five times more data was going to be going through data centers and using electricity.

That that may or may not come to pass. That's a prediction. So there are some who say that is, and some who say that isn't going to happen. But it's a huge amount of electricity, and it's a huge amount in terms of carbon footprint. And some estimates go between about 1% and some say up to four to 5% of electricity is...

Jeremy: In the world?

Paul: In the world.

Jeremy: In the world, right.

Paul: Yeah. And if you take that in terms of carbon footprint, we're talking in the magnitude of somewhere between something like a quarter to two to 3%. It's a huge amount of actual carbon emissions that go into data centers.

Now there's an interesting conversation about what constitutes a data center at this point, because some people go, "Well, it's only hyper scale computing." And you're sitting there going, "Well, actually that limits what you mean." And it's a whole other conversation.

But I tend to put that it's actually around 2%, probably a little bit less, which puts it somewhere in the same region as aviation pre-2020.

And so you end up with this... Even if the numbers aren't right, they're in the same kind of ballpark. And it's really actually very difficult to know how much electricity data centers actually use, because most of the providers don't tell us. And a lot of the people who do the digging around the numbers are doing things like going and counting the number of diesel generators outside to work out what the backup power is, and therefore how much backup power you need for the size of... For a type of data center and what that would...

So they're not, nobody really knows, is the actual answer. But it's actually significantly more than most people realize. And while there are some that go, "It's really, really terrible and it's huge," and there are some that go, "It's nowhere near as bad as all that and it's coming down," and it's probably somewhere in the middle. It's really not good, however we skin this cat.

And so we end up with data centers being a problem, and they are an issue we need to solve. And actually, if you look at someone like Google, and if you look at someone like Microsoft, they're both trying to do an awful lot in this area.

Jeremy: Yeah. Well, that's what I was going to say. So I mean, you do see press about this and you do see companies talking about this. What are they doing to try to become more efficient?

Paul: So if you have a look at, let's take Google for example, because I think they're probably the best example of good practice in this space. They are trying to look at their electricity usage. They're trying to offset on a, I think it's an hourly basis. It might be an half hourly basis. They're trying to offset their electricity usage usage with purchases of renewables in the same grid. So in an electricity grid, they'll buy. If they using X, they'll try and buy X amount of electricity from renewable sources for the same hour in the same grid. So there are some places in the world where that's not possible. So some of the, I think it's the Chinese data centers. I think it might be Taiwan or something, they can't do that because there aren't the renewable sources available. So they have to do it in another way.

But a lot of their US data centers, they're definitely able to do that for the majority. So that's a really, really good practice, but they're still not able to do it completely. So they're still offsetting in other ways. And offsetting here is, we use... You use the electricity in the grid, which may not be completely 100% renewable. And then you buy an amount of renewable energy from somewhere else to offset the fact that you've used bad energy, carbon emitting energy. That's all an offset is.

Microsoft again, they've got interesting practices here. I love this topic, by the way. They use renewable energy for something. I can't remember the exact numbers. It's something like 40% of their data center usage. So that's pretty good. And then they offset the rest, but they also have an internal carbon price. So if they emit, then they have to pay internally. I think the last time I looked, it was 12 or $20 or something like that. It may have gone up since then. So for all the carbon they emit, they actually price it internally, which means that if they move to renewables, it's actually cheaper for them.

So it's actually really, really interesting things going on at Microsoft, and I think that's an inte resting model to look at. AWS...

Jeremy: It's a good way to play with... A good way to play with your balance sheet is to do that.

Paul: Very, very good. Really good practice. AWS, they have four green regions. So there's Frankfurt, Ireland, Montreal, and Oregon. Really good, except for the fact that they don't tell us how green they are or how they're offsetting or what they're doing there, but they just tell us they're green. So it's a little bit unhelpful. And they don't tell us how green any of the other regions are. So you can't actually offset your AWS bill, which doesn't really help if you're... I'm sure if you're really big, you can go to them and say, "How much electricity are we using at these different points, and can you tell us how to offset?" And they'd probably do it, but you have to be kind of... Hundreds of thousands.

And if you're serverless, the likelihood of you doing that is very, very low. So I think you see that there are interesting conversations going on in that space around how best to start thinking about the carbon footprint of data centers.

Jeremy: Right. So it's not just the carbon footprint and the offsets, and some of that stuff is... I think those are great, right? I mean, you're still using dirty energy, I guess if you want to call it that, but you're subsidizing essentially green energy on the other side.

But is there a difference or a distinction between sort of the idea of just offsetting your carbon footprint saying, "Hey, listen, we're going to run up the meter, but we're going to pay for it in another way." Is there a distinction between that overall environmental impact and the efficiency that can be created in a data center?

Paul: So I think efficiency is a really difficult area for a lot of people. Because a lot of people like to go, "Oh, we've bought better servers and they're more efficient, therefore it's better for the environment."

Jeremy: We installed LED light bulbs.

Paul: Exactly. We've done our bit. There's the thing called Jevons paradox. It's a very old economist who's long dead who basically worked out that if you made things more efficient, made energy production more efficient effectively, then people would use more energy.

Jeremy: Right, exactly.

Paul: So, you basically make it cheaper to do something, people will spend more money on it and therefore they use more in the end because it's like, "Oh, it's fine. This is easier."

Jeremy: And not to not to interrupt you, but that's actually the same argument people make about serverless, where they say serverless is faster and cheaper. That just means you'll build more with it. Right? So you're not actually spending less. You're just doing more, which is still great. But back to your point.

Paul: Yeah. And it's exactly the same with cloud. When we all had to buy servers and stick them in a data center ourselves, that was hard work and it was difficult. And then along came the ability to buy a virtual server. And then we literally just bought virtual servers like they were water. And then it was like, "Oh, it doesn't matter. It's tens of dollars and it doesn't matter."

And now we just go like, "Right, we can build whatever we like." That's efficiency. And the fact that a data center is more efficient does not make it more environmentally friendly. It just means that everybody wants to use more of it. And that I think is a... I think it's a fallacy that if we make things more efficient, it's more environmentally friendly. Yes, we need to make things more efficient, but we need to make things more efficient and we need to look at reduction of our carbon footprint overall.

And I think that reduction comes first, and then efficiency leads to reduction. Not efficiency is just the overall goal, because that doesn't lead. And that I think is a personal concern as well. Because most people talk about, "Oh, it's fine. I'm becoming more efficient. I'm personally, I'm doing all these wonderful things."

And then you sit there and go, "Well, that's fine. I I'm doing my bit. We're more efficient. We don't do this. We don't do that." And then you're like, "Well..." But then you use more heat because your house is... You just do and you just use more because you've made yourself more efficient, or you...

A good example, I think is, most people would understand right now is in the middle of having spent three months in our homes, which is when we're recording this, you all of a sudden realize that you can have packages delivered to your house.

Jeremy: Right, it's so easy.

Paul: It's so straightforward. The efficiency of packages being delivered to your house. Stuff, you just go, "Well, I don't need to go anywhere. I can just order it. I can just order whatever. Well, I just need more of this. I will order it." And it's just the efficiency of stuff. Doesn't mean you use it less. It just means that you use... You just find it more and more. You will use it more. And I bet everybody who's listening to this is going, "Oh yeah, I've ordered far more stuff in the last few months than I have in the previous two years." And that's how it works.

Jeremy: Well, I also love the fact that you get that Amazon box delivered and it's a box within a box sometimes within a box, right? Like, I mean, it's efficient for you, but it is a lot worse for everything else in terms of how many boxes need to be made. Which by the way, I had this idea, I don't know if anybody's on board with this, but when you get a package from Amazon, now that Amazon is doing most of their own delivery or at least in some places, you should just leave your broken down boxes out on your front porch or wherever it is. And then the Amazon people should take those back and then reuse those boxes as opposed to trying to throw them into recycling, which I think most towns say they have recycling and then it ends up in a landfill somewhere. But anyways, separate idea.

Paul: Whole other conversation.

Jeremy: Right, exactly. So you mentioned though, this idea of again, efficiency probably creates more demand in the sense, because it's easy. Right? And so if it's easier to do, then again, you can consume more. And I think that is certainly true with things like EC2 instances. Like you said, I can just spin up as many EC2 instances I want to, and it's going to cost me incrementally, but for the most part, it's pretty cheap. Whereas when you build with serverless, you kind of build with some constraints. So does that tie into green energy or, I guess, maybe fighting against Jevons paradox there?

Paul: Yeah, I think it does actually. I think there is more in there because your constraint is you're trying to reduce, you're trying to use as little as possible when you're trying to build serverless, or at least in the way that I built with serverless. You're trying to say, "How can I do as little as possible work in terms of compute and get as much as possible out of it?"

And I think that is the most efficient thing you can do. And it's like your end point is, with all of these efficiencies, is to get to the point of only doing as much as you have to, to achieve the goal that you're trying to get to. I think if you're using something that is inefficient, then you always have efficiencies to bring in.

I think if you're trying to build something that is as efficient as possible and there is pretty much no... There's nothing to make more efficient in that process, then you're essentially building as green as you possibly can, whatever the goal of the solution at the other end at the end of the day is.

I mean, just as a quick example, one of the principles that I had when I was building the startup in 2015, '16, was that we didn't put any... We were using Node and we were using Python as our libraries within AWS Lambda as our languages. And we basically said as a rule that we use no libraries. We use no libraries, we use no frameworks, we use nothing. It was all pure language in each of the functions. And each function should only do one thing.

And it was like, the principles meant that each function, I think we only had like three or four functions that were over 150 lines ish, maybe 200 long. So if you went into a function, every single person in the company who was able to code could go in and probably fix the function if it went wrong. And it also meant every single one was lean to within an inch of its life, do you know what I mean?

It was that kind of, it was so efficient and it was so clean that it was getting ridiculous. And it was easy to do because it was just the principles that we set in place. It was the constraints and the principles and everything else. But if you don't start with those principles, if you don't start with the understanding that that is the constraint and you just go, "Right, we've got all of this, all of this. We can do what we like within the context of the world. Oh, isn't this lovely?"

If you don't start with those understandings and constraints, then you end up in, "Well, how do we make this more efficient? Can we make this more efficient?" If you start with, "How can we be as efficient as possible," then it's a different conversation going forward. And I think that's the serverless...

Jeremy: Yeah, no, I think you're totally right. And I mean, that's one of those things where, like you said, the single purpose function thing is... Not only is it great from an efficiency standpoint, you're using as little, you're spending as little money as possible because it loads fast. You don't get the cold starts as badly and things just run really quickly.

It does one very specific thing and it can scale independently, right? So if you have millions of people hitting against that one thing, that one action scales, whereas something else like maybe your delete AP, your delete path or whatever doesn't have to run. That doesn't have to scale.

And it's not like a microservice that's running on a container where that container is running and have to scale up everything just to process one small bit of code. So I think that's really interesting. And I love this argument about thinking about efficiency right from the beginning. And you and I have talked before about where that efficiency comes in later on and what that means for global impact or for climate impact later on.

But I want to talk quickly about Project Drown, sorry, Project Draw Down because you talk about this a lot. And I think this is a good way for developers who are unaware of their impact to start thinking about this. Can you tell people a little bit about that?

Paul: So I think there are a lot of people think that to fix climate change and to kind of hit all of these goals, the obvious thing is we've just got to sort out the electricity system. One of the obvious things we've got to do is go and just make everything renewable electricity. And they don't know things like that 33% of global emissions come from agriculture, or that cement is a significant emitter of carbon emissions or that the oil industry itself is like 10% of emissions because of the fact that actually taking stuff out of the ground that is going to be burnt and emit emissions actually produces emissions itself. It's a massive, complex world. It's not just about this we need to make more batteries and the world will be a better place. The complexity of the argument about how we fix the problems that we have are huge and systemic. And most people haven't actually looked at this problem to know about it. And I think Project Drawdown is brilliant because one of the things that we almost certainly will have to do to hit the kinds of targets we need to hit to make a livable world in the future is that we're probably going to have to take carbon out of the atmosphere in some way, or at least reduce, significantly, the amount of emissions in certain areas.

So one of the organizations that looks at this is something called Project Drawdown. And they actually produced in April I think it was an updates to a previous piece of work, which was looking at the biggest impacts in what kind of things will reduce carbon emissions significantly. So what will effectively draw down emissions over time as quickly as possible. Previously, it was things like air conditioning units, because actually the stuff in air conditioning units is impossible to recycle. But actually air conditioning units, when the planet's getting hotter, we're going to need air conditioning units. So if we produce more air conditioning units and the planet... It's just not good. So we need to do something about air conditioning.

Jeremy: It's a vicious cycle.

Paul: Brilliant. Yeah. And, but previously they were also talking about things like educating women and actually because educating women is important for understanding how to, across the world, this isn't in various of place, but across the world in terms of understanding, because they see that the more educated that women are in terms of in areas where they aren't educated, there seems to be greater population. So this just seems that this is a good idea. So creating organizations to go out and educate women is a climate. It's a good thing for the climate. You're like, "Okay, this is a positive." So you see that it's not just about going and building massive wind farms, and it's not just about building batteries. And it's not just about getting everyone to drive electric cars, although those are also good. And it's also things like agriculture, and it's also things like lab-grown meat. And it's also things like all of these elements actually are a part of the solution and it's understanding what are good things and bad things to do and understanding how they all fit together. And I think it's worth just Google Project Drawdown, it's worth having a look at their website, reading all the stuff around it. I think it gives people a more rounded understanding of what's needed to be done rather than just going, "It's just this."

Jeremy: Right. No, actually I was taking a look at it after you and I talked before, and just the fact that like family planning, just, that has a huge impact on carbon emissions. You know what I mean? And just everything from the medical services that are needed to actually having another child and what that means and just being able to understand those impacts. So there are obviously those big things that you can do. These are global things like, yeah, let's get rid of fossil fuels and some of that stuff, but what personal actions actually matter when it comes to... So we'll move away from technology here for a second. Because this is an interesting subject for me. My wife and I talk about this all the time.

But what personal actions actually matter. And because the other day, and this is something silly, but we recycle or we try to. We put the stuff in the recycling bin, hopefully it gets recycled. We try to buy aluminum or glass instead of plastics and things like that. Recently, we just switched because I was getting sick of throwing away all these paper towels and napkins. We just switched to cloth napkins. Now, again, does that have a huge impact? Probably not. Does it make me feel better about myself, a little bit. But what are your thoughts, because I know you're a vegetarian, like what do these even mean? Does this help?

Paul: Yes, I think it does. I think there are two things that's worth saying. I think all these things are worth doing. I think they are worth doing from the point of view of changing who you are and changing the way that you live. I think those things that are going to be important, I think over the next 10 years, the world is going to change. I think we are going to see a radical shift into something looking like a low-carbon economy. And I think a lot of people are going to struggle to shift over to that. And so I think the more that we can do to understand that and understand that we are probably going to be eating less meat, that we're probably going to be reusing an awful lot more of our items, that we are probably going to have less access to some of the things that we see as convenience items.

We probably will still have cars, but I suspect that fuel will be significantly more expensive because it will be taxed, not because the fuel will be unable to be gotten. It's just that I can see an awful lot of things. So I think we need to think about our own personal lifestyles in the context of the world is going to be changing. And I think it's probably a good thing to start considering how your life needs to change around all of those things. So I think it's important to think about those things and understand that actually we do live an incredibly privileged life. We talk about switching lanes a little bit, we talk a little bit about how we have... Talking about Black Lives Matter.

Being white, I literally don't know, well, that I have privilege simply because of the fact that I am white. It's just, it happens. And I don't know. And I've learned an awful lot about all the little things that I've never had to worry about. And I think if you switch lanes back, we live in the industrialized nations. We have had the oil and it's a similar kind of thing. We have had the money and the wealth and the oil that. And that other nations who haven't had the money in the wealth of new oil haven't had. And I think we need to realize that that isn't going to last forever. That tap is going to get turned off. And I think the wealthiest nations are going to struggle the most because they are going to have to change the way that they think.

And I think that lifestyle, the idea that the green economy is going to change all of that. And they're just going to be able to carry on and keep staying rich, I think isn't going to keep going. So I think in terms of going back to the personal because that's where it all started. I've just gone on a bit of a rant. I think you and I, and all of us, I think we all do need to start thinking about the personal. I think we need to change those things. Recognize that they aren't in and of themselves going to change anything in a big way, but getting involved in a movement is. So in the U.S. it'll be writing to whoever your representative is or ringing them up or whatever it is.

I see all these things from the U.K. and think, "Well, we don't do that. We just write to our MP," and then that's about all we can do or go on a march. But it's those kinds of things. Finding the groups that actually talk about these things and understanding the intersectionality of climate with other areas of justice, including race. And I think these things are quite important to understand that there is a connection. For example, it's something that I think is quite important to point out is that is an awful lot of non-white. And I think it's worth just saying outside of white people, non-white people live in more polluted areas simply because they are poorer. So there is a racial element to climate justice. And I think that you can't just separate these two things out. So I think you can't just go, "I'm just going to buy better things." I think there are other elements to being part of the conversation that just look into it and see where it fits.

Jeremy: Yeah. And I think you're right. I think everything is so intertwined and complex. And I think that goes back to something that I thought about for a very long time where I feel like it's the infrastructure that is part of the problem. Like you said, we've got oil and we've got combustion engines that we use to power our cars. We have highways that are terribly inefficient in terms of backing up traffic and cars idling and things like that. In the United States, I've been to Europe a number of times and in a lot of the cities in Europe, they have very efficient public transportation systems. In the United States, we don't. I live about 45 minutes outside of Boston, Massachusetts. If I drive my car in during rush hour, it takes me an hour and a half.

If I take a train in, which is just an old clunky commuter train or whatever, that takes me an hour and 15 minutes. So the trade off of the flexibility and so forth. And again, I'm probably not saving much in emissions either, but just going back to the infrastructure thing, this is something where if I'm sitting at my house right now, like you said, I'm privileged, I live in this nice house. I've got central AC, I've got heat that I can just turn up to as high as I want to and as be as comfortable as I want to burning oil. Because that's what was built in this house. That's what I have. I have oil heat. The question is as a consumer, and this is super selfish, but why do I have to be cold? Because the government and because corporations, and because everybody made these decisions for me to say, "This is how you get warm and you get warm by burning oil," as opposed to saying, "Hey, we've covered the Texas panhandle with solar panels and we're collecting all this energy and we're going to have electric heat it's going to be a hundred percent efficient and you can turn the heat up as much as you want to and that's not going to affect anything."

Now again, I know that's a little bit selfish, but I think about these things just from an infrastructure standpoint to say, "Why do I have to make those trade offs?" And I think a lot of people would ask this question is like, "Why does my house, why are modern houses built with one waste pipe? Why isn't there a gray water and a waste pipe? Why aren't those separate? Why can't I capture that gray water and use that to water a garden or something else? Why aren't houses built that way?" And so I think that's just this thing that bothers me so much is that asking people to do a lot of these personal things, and I have no problems doing these things, I certainly don't have problems doing these things, but I feel like the infrastructure is fighting against us.

Paul: Yep. And I would agree with you. The very simple answer that I think, but yeah, the slightly longer answer is that oil companies, it has been shown in various different ways. And I think it's worth pointing to other resources on this, have spent an awful lot of time, money, energy resources in making sure that they are the ones who make the rules. And they are the ones who have the power to do things. And it's like there were electric scooters in the 1920s. So people think you get an app and then you go and you scan the QR code, you pick one up and you scoot off and isn't that wonderful. We had electric scooters back in... we were doing stuff like this. The first cars. There was a discussion whether it was going to be electric back in the day. We have an oil industry that is incredibly powerful across the world.

And oil is seen as incredibly cheap, and it's incredibly efficient in terms of you burn something and it produces an awful lot of power and energy. And so there are some really positive things about that, but the downside is that it produces the emissions and those emissions have a secondary effect that is incredibly negative. So we need to move away from that. It is definitely a systemic problem. It is definitely a systemic issue. And it is definitely something that you can't fix on your own. So part of this whole conversation about people going off grid, and I've seen an awful lot of people in tech, I've watched them, they're going, "Yeah. I'm doing my bit, I've got loads of solar panels. I've got a heat pump in my garden. I'm doing this, I'm doing that, blah, blah." And I'm sitting there going, "Well done. You're going to be fine while the rest of us are absolutely stuffed." And it's just like we have to kind of think about all of us. And all of us is politicians. It's companies. It's understanding that a pledge to do something by 2030 by a company is not the same as we've done this this year. And it's just thinking about those kinds of things is far more important. And putting pressure on companies, I think is also quite important.

Jeremy: Right, now, are you a fan of the Back to the Future movies? Have you ever seen those?

Paul: Yes.

Jeremy: All right, so, because I'm just thinking that if electric cars would have been a thing, then the entire plot of Back to the Future III would have been unnecessary and they wouldn't have had to make that terrible movie. So, but anyways.

Paul: It would have been different.

Jeremy: It would have been different. So one of the things though that I guess... now that you and I, we've solved the world's problems here because we've discussed them on a podcast, obviously there's a lot that needs to be done. But practically, what can people in tech actually do to try to have an impact on this?

Paul: That's a massively broad question. So right. I think the biggest thing is make yourself aware, start to do the research. A really good place to start is, it's something that a lot of people have heard of, but probably not read, is just to read the IPCC, Intergovernmental Panel on Climate Change, Special Report of warming of 1.5 degrees, the special report at 1.5 degrees, it's fine. See IPCC. It came out in October 2018, but it was the thing that triggered most people into understanding that we had not a huge amount of time left to really try and keep warming below 1.5 degrees. And that was where the idea of a carbon budget came from. The amount of carbon we had left to burn.

The amount of carbon we had to put into the atmosphere before we basically breached 1.5 degrees as a threshold, or two degrees. And so that was where the idea of 12 years came from. And so all of these little things came from that paper and it's well worth reading. So I think that's worth kind of just pointing as a good reference document for all of this. But in terms of tech, simple, start looking at your data centers, where you store stuff. Are your data centers run on green electricity? If they're not, go and ring your data center provider and say, "Move yourself over to green electricity, please. Thank you very much." It's really not hard and actually half the time it's cheaper. If you're in the cloud, move your workloads over to green regions.

So if you're already in Google or Microsoft, you're probably already in a green region. If you're an AWS, move them over to Oregon, Montreal, Frankfurt, or Ireland. If you can't move them over, because you're basically stuck, which is often what happens, then move your workloads that can be moved. So your things like your test environments or your machine-learning workloads or things that don't matter which region that they're running, move those over to green regions. There is always a way of moving more stuff over to green regions. Turn stuff off is also something else. Don't leave servers running overnight. It's all of these little bits and pieces that people are like, "I don't understand." Why do people just go, "It's fine. It's just it doesn't matter. It's over there. I don't need to think about it." Turn computers off overnight. Your own work unit. These are habits. Once you understand that, it's just simple things to do.

That's kind of important. There was another story I read, was it last week in Wired, or it might have been this week, about a guy who has a WordPress plugin, and the plugin, he looked at it and he realized that if he just cut down some of the code, and like 2 million people use it, but just cut down some of the code, he could reduce what the download speed was, the amount of download. And he's worked out it's around, I think he said 57,000 kilos of carbon emissions saved just by doing that. I think that's off the top of my head, but it's like it doesn't take a lot to start to think about how can I reduce the impact of what I'm doing. And even to the point of going, "Can I just reduce the amount..." Taking libraries out of code. Taking frameworks that you're not using out of the code. Actually just reducing the amount of code you write is just a simple thing to do.

And it's not that it's going to save the planet if you take one line of code out. Well done. It's just the principle. It's just following through on a principle. And it's just taking these principles and saying, "This is what I'm going to do ongoing." But that's all it is. Is it's a set of principles that if you can stick to them, then after a while you find that you just keep doing them and keep doing it. And you are finding that it gets better and better over time.

Jeremy: All right. It's the straw that broke the camel's back, but in a good way, going the other way. Like every little bit helps. Yeah, and I think that's something that is a powerful motivator for startups is saving money. So trying to be efficient in those ways. And I know I've worked for a lot of bootstrap startups or a lot of startups that were like, "Let's keep the cost down." So I think, again, working with serverless, having those constraints, having that mindset of we need to be more efficient, having the mindset of we need to save some more money. These are all benefits that really should benefit your company in the long run anyways. And I guess that's another thing too. What about organizations? So what can you do? You're just a staff engineer at a company, what do you do to try to affect change at your organization?

Paul: So this kind of goes back to the work that I've been doing with the Leading Edge forum up until, well, I'm still in the middle of doing it. It's kind of working with them on looking at organizations and how they change and grow and look at climate and look at sustainability as a wider goal. One of the things that is clear is that actually organizations don't know things like where all their emissions come from. They don't know things like understanding whether their supply chain is full of good or bad emissions. They don't know these things. It's actually quite difficult to understand. So it's actually a simple thing to go and talk to your head of sustainability and go, "Well, actually, what is the situation?"

And I actually, I know this, I've done this at a number of companies. If you go and talk to these people, they love talking to technical people because then they can go, "Can you give us a hand with this complex technical problem?" And you're sitting there going, "Brilliant. I've now got some things that I can..." It's like, "They want something." And over the next few years, I think we're going to see a connection between data and sustainability that is going to become a quite important key thing. So I would suggest that if you're just a staff engineer at a small company or something like that, I think putting yourself into a conversation with head of sustainability or asking who is leading on this, if you're in a smaller company, I think it's actually quite important.

I've talked to a number of smaller companies who, they've asked me, "Well, how can we reduce our carbon footprint?" And actually, that's a very difficult question if you're a smaller company. It's like, well, this is pre-2020, but they were doing things like flying to conferences and understanding what can we do then? It's like, well, either you reduce the amount of your flying to conferences or at the very least you offset and offset more than you would normally offset, but you need to understand reducing is actually quite important. So think about what you're doing there. So maybe think about doing remote conferences and remote videos and being a leader in front of that and talking about why you're doing that. And a couple of companies have taken that on board and started to do that anyway, before all of this.

And I think we might see more virtual conferences anyway after this whole change in the way that we do things, I think people are becoming more used to doing virtual work and virtual conferences and seeing the value in all of that. So I think it's just starting to become aware of, again, all those little things. It's like, "Well, we use servers, we do this, we do that." Talking to the senior people and actually starting to ask questions, "What are we doing in this company? How are we doing? How can I help?" It's just a really easy question. And you'll find that actually not many people are actually asking that question. Most people don't. A lot of people who are, even in large companies, even in very, very large companies, they don't necessarily ask that question. They want to do something, but they don't necessarily go up to someone and go, "Well, what can I do?" And I think that's a really straightforward and simple thing to do. And people will respond.

Jeremy: Well, you made a good point about the travel for conferences and things like that. And I know you were one of the founders of the ServerlessDays Conference or JeffConf originally, which I think is an amazing thing. I really like that idea of regional conferences, in the sense where people can take public transportation or they don't have to fly to get to these, and then you bring in some speakers that do fly, but it's better to have 10 people fly than to say, have 65,000 people fly to Vegas for re:Invent or Google Next or some of these other, really, really large conferences. I like those small intimate regional conferences because again, those are more efficient, I guess, than making a lot more people travel to one place as opposed to having a few people travel, and it's all easy to get to.

All right. We talked about the big three a little bit and compared them in terms of their green and stuff like that, but in that paper that you wrote, you have this cloud league table in there where you compare them. I'd love to know more, what about Alibaba and Oracle and IBM and some of these other things, where do they all stack up against one another?

Join us next week for Part 2.

This episode is sponsored by
Amazon Web Services and Epsagon