DEV Community

Serverless Chats

Episode #60: Going Green with Serverless with Paul Johnston (Part 2)

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/DTpP7RGXV6g

Transcript

Jeremy: 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?

Paul: They aren't as big. Let's just be clear on that one. They aren't as big, and their green credentials are less clear. For example, Alibaba is very big in China for obvious reasons, it's a Chinese business and yes, they have a footprint outside of China, but they're a primarily Chinese business. When we looked and researched and were trying to find out about all of their green credentials, we found very little information whatsoever. It was almost non-existent. We found a little bit about efficiency in data centers and putting things in cold regions of China. You're like, "Well, that doesn't actually change anything if you're growing at a massive rate." It changes the conversation a little bit.

Jeremy: Can you do that though? Can you pack your servers in snow? Does that help with the cooling bill?

Paul: It depends on the server, I suppose. But I'd say you end up with this, not every conversation is equal. This is shown across the political spectrum as well. The conversation in China is very different to the conversation outside, in terms of industrial nations. Industrialized nations such as the U.S., UK, Australia, and Europe as well. You have very different social context. But anyway, coming back to Alibaba, we just found very little information. IBM and Oracle, actually, both had lots of information. IBM has had a commitment to renewables and sustainability for a very long time. I think since the '70s if I remember right. They have good credentials, but they don't offset their renewables in data centers or regions.

Paul: I think I remember, from the top of my head, because I can't remember everything, they are renewable in the UK. I think Oracle are renewable in the UK anyway. But some of these regions, they are renewable, but not all of them. But it's not clear unless you dig into the paper and unless you dig into their information. Nobody, as far as I can tell, has a little green dot against a region that says, "This one's renewable." It would be so much easier if they did. These other organizations, none of them, apart from Google and Microsoft have really made a play for being the green advocates in the space. Amazon does, but that's a whole other conversation which I'm not going to go into at this point. But the three lower down, I think, are struggling to be able to play in the same conversation. I think they would like to be seen as green, but I don't think they are really pushing the agenda because they don't see it as a point of differentiation.

Jeremy: All right. Then what does the tech industry have to do as a whole? I know you had some recommendations in your paper about this, but just what are maybe the top two or three things that the tech industry as a whole could do to address the climate crisis?

Paul: The climate crisis as a whole.

Jeremy: Or I guess their impact on it anyways. Let's start with that, we could build from there.

Paul: I don't know anymore. I think I've gone backwards and forwards on-

Jeremy: Don't give up Paul, don't give up.

Paul: It's not that. It's just there are so many things. I think my biggest thing is the tech industry needs to find its activist voice. I think that would be my point. I think sitting there and going, "Oh, everything's going to get fixed by technology," is entirely the wrong approach. I think my personal view, as much as I like Tesla's technology, I don't think Tesla is going to save the world. I'm not an Elon Musk fan, I find him very difficult in a number of different ways. That's as much as I'm going to say, but I find-

Jeremy: I don't think he listens to this podcast, so don't worry about it.

Paul: That's fine. But I find a lot of people within tech look at techno utopianism, and let's call it that because I think that's a pretty simple way of it. That technology will save us. The more that I look and the more that I look at what is happening in the world and the speed the technology is evolving and what we need to do in the speed that we need to do it in terms of climate change, I don't think we're going to get anywhere near fast enough technology evolution. We have to do something else. We can't just expect technology to catch up, fix it, and just for us to carry on.

I think technology needs to learn to find its activist voice. I think we need to be activists against those organizations who are not doing enough, who say they're green and are not, and in this, Amazon, I will call this out. Much as I love their serverless technologies and the people who work there are brilliant, the wider organization I think is not doing a good enough job in terms of its green credentials. That hasn't been good enough from my point of view. I want them to do better. Because actually, I think a green Amazon would be a great thing for the world and I think they can do it. I have seen Amazon and what it does when it's amazing, and I think if they turn themselves around and actually did the green thing properly, then I think that the hope for the future would be significantly higher.

That is why I want Amazon to change is because I think they have the power to be a force for good, and I think they're not doing that at the moment enough. That's one. I think find the activist, find the place that you within tech want to change and go and change it. Because I don't think we have anywhere near as much time as people think. I think your career in tech in 10 years time will not look like the career in tech that you think it is now. We are in a completely changing environment. Depending on what happens in the U.S. in the next few years, the world is changing around us. We are in an inflection point that I don't think many people are aware of. This is just slightly rounded conversation, but in terms of people in tech, it is very much, don't just let it happen around you. If you want to make a change, go and do something about it.

To that is, green your data centers. It is go and find an organization to go and get involved with. It is find your political voice, whatever that is. It is inform yourself. It is get involved with all of the things like all of the black lives matter protests and all of that. I think it is important to find all of those areas and get involved, because once you start getting involved, you will find other areas of intersectionalities that will then help you look at it and see all of the wider issues and all of the systemic problems. And then you will start to go, "This is huge, and we've got to do something about it." It just becomes a bigger problem.

I think the tech industry needs to, and I know this is a massive rant, but speaking as someone from the UK, I look at the U.S. and I see the U.S. tech industry, and I see the money in the U.S. tech industry, and I actually think that I don't think we need the technology, I think we need the money and the brains to go into other areas. I think we need them to stop thinking about how to make technology, I think we need to start them thinking in how to start being humans again. There's a very big difference in that. I wrote a blog post about it, it was quite fun.

Jeremy: Well, I don't even know where to start to respond to everything that you just said, other than I agree with you. Brilliant. That was great. The funny thing you mentioned about Amazon just not doing as good a job as it could, I think that is again, partially an infrastructure problem, in a sense, partially a priority problem, partially this thing where I think it's FedEx decided to stop delivering their packages so they had to speed up their own fleet for delivery. Why not electric delivery vans? The technology's there. Maybe it would've cost a little bit more, maybe it would have taken a little longer to roll out, but those little things, I say little things, huge things like that that they could have done that would have had a massive impact.

Again, singling out Amazon is probably not fair. I think every company in the world, the vast majority of them are under that same thing. Any incremental change is good, but I think you're right, we just need massive systemic change at this point. Otherwise, that clock is ticking and we're going to run out pretty fast. I guess another question though about that is big companies like Amazon and Microsoft and IBM and even Oracle and some of these other ones, they have resources. Big tech companies, Facebook, Google, the Twitters and things like that, they're building their own data centers, or of course they're building data centers that are shared, that other people can use. Are they more likely to become green or have more efficient data centers and be more up to date because they're not...

I think about when I rented a rack in a co-location facility when I had my web development company. I had one rack with the power coming in, whatever, I had servers in there that were six or seven years old. Because you're like, "Some of these old websites running on that thing, I'm just going to keep it running, and it was probably terribly inefficient." It's a massive investment for companies to continue to upgrade their servers and continue to make their data centers or their on-premises locations green. Is that something where maybe there would be a nice point where moving to the cloud actually would be the smart move from a green efficiency standpoint?

Paul: It's a massive question.

Jeremy: Sorry. Well, you had a massive rant before that. I'm just getting you back.

Paul: In the end, I know the joke is cloud is just other people's servers and all that kind of stuff. It's always underneath it. There's just servers and there's just servers. But I think that trying to make these servers more efficient, trying to make these data centers more efficient, there is still constant churn. We don't keep things efficient. Two, three years down the line, the server that you were using is not efficient. Six years, seven years, it's old. You don't want to be running stuff on there, you want to be running stuff on something that's efficient and new. Actually, there's an enormous amount of e-waste in terms of the data center industry. It's not straightforward.

The conversations around all of this are not straightforward. I think everyone needs to start thinking about moving to the cloud simply because we need to be reducing our impact. If you're running stuff, I think it's important to be able to go, "Actually, we need to be able to reduce the amount we run." But that means, understanding how that cloud, that you're choosing to work with, is working in terms of its sustainability. You can't just go, "We'll move it to X cloud, or Y cloud or Z cloud, or whoever it is, but we'll trust them to do the right thing."

You've got to still have that relationship. You've got to still be able to go that cloud, "You, Mr. or Mrs. Cloud person, you've got to tell me, are you using green electricity? Are you using renewables? How are you disposing of everything? What is your supply chain?" I think that conversation over the next few years is actually going to become a much more common conversation. It's going to become more important. You are not going to be able to get away with, "We just run efficient data centers." That's not going to be the standard and reasonable response. That's going to be a table stakes. Green data center will be a table stakes conversation, and the best practice will be, "Well, we're actually running 100% renewables and we're putting more into the grid, and we're being as good a partner as we possibly can. And all of that. We haven't got diesel generators, we've got batteries."

It's all of that conversation that I think comes back to. Maybe we will end up not using certain companies because their data centers are not green enough. Maybe that is where we end up, that actually societal pressure actually pushes these companies to do better. But I don't think we're there yet. I think we're probably a couple of years away, two, three, or four maybe, away from that.

Jeremy: Right. I think that that conversation about e-waste is probably really important too. Because now I'm wondering, where did my Pentium 166 megahertz computer go from 25 years ago, and my 32 megabyte RAM module? Is that a landfill somewhere? Do they melt it down? Is that in my new MacBook? I have no idea. I think that's interesting what you said, that again, you want to make sure that the cloud infrastructure is efficient and it's green and you want to do that.

But I think there's a bigger conversation around this as to say if I was to buy, I don't know, a million dollars worth of Dell servers, brand new, highly efficient servers. I throw them into my data center, I'm running my own power, I have to buy my own batteries, I have to have my own backup, I have to have all of the waste that's involved with that. Even if I make it incredibly green, two years from now, there's faster chips, there's new servers, maybe I can swap out the chips in the servers, I don't have to get rid of the actual metal, the casings and things like that, but I have to make another massive investment in order for me to then upgrade.

Whereas if I put my stuff in the cloud, even if the cloud is not as green as I want it to be today, maybe tomorrow they're a little more green. And then the day after that they're a little more green. And then five years from now they're 100% green and it didn't cost me any more money other than what it cost me to run my workload. I go back to this, and I don't want to offend anybody here, but climate change deniers, people who don't believe it's happening, and I know we had to change it from global warming to climate change because they're like, "Oh, it's actually getting colder in certain areas." You're missing the point.

But my thought here is to say, is there anybody who disputes the fact that if you dump oil in a pond, that that's a bad thing for the environment? You don't want to drink that water, you don't want to swim in that water. Can we agree that pollution is a bad thing. Whether it's making the earth hotter or whatever, can we agree it's a bad thing. If we can agree on that, then that's a good thing. That gets us where we can think of maybe the moral part of this. But let's take it back to a more selfish level. If you are a business and you can implement green things that are going to cost you less money over time, I use serverless. I spend, I don't know, a 10th of the time writing code than I did before. I have teams that are smaller so I don't have to have 20 people to write an application. Now I can have five.

The impact of that on the planet is huge, but also, the impact of that on my wallet is huge. I'd love to get your perspective on that because I think that there is a snowball effect, that as you make one small change to increase efficiency, to reduce your carbon footprint, to do these things, save yourself some money, but then the impact of that is huge.

Paul: I think that you make a very good point. Going back to 2015 when I was CTO of a startup and it started off with me building serverless stuff with AWS Lambda and DynamoDB, we were doing half a million monthly active users with a backend team of three. We weren't building complex servers and putting servers in a data center and having to run that and having to think about automation and dev ops and all of this kind of stuff, and having to think about end of life of whatever and all of that kind of stuff. We were literally able to change with whatever we thought we needed to worry about, and we could have done it in seconds if we actually needed to. We could have changed the way.

That was the architectural choices that we made in terms of the technology and moving to the cloud. I think that makes a huge difference. We were able to do things like be all remote from the very beginning. Being remote means that we are not traveling to an office, and when we're not traveling to an office there aren't the carbon emissions from traveling to an office. Yes, we're using electricity, but we can all use green electricity at home. That's a heck of a lot easier than having to think about whether or not our train or our car, or everyone buying electric car is ridiculous because that's quite expensive.

All of those knock-on effects of having a smaller team, having a remote team, thinking about smaller workloads, which takes less electricity, all of those additional elements, it's not just making a technology choice. As you say, it is that continual change. That comes back to the original point that I was making, I don't know, an hour ago was it? A while ago anyway. Which was that-

Jeremy: It's been awhile.

Paul: It's been awhile. But that constraint, that you put that constraint on yourself of how can I, in making this application, in making these decisions, reduce my impact as much as possible? Things like going remote, it was 2015, it was a no brainer. I didn't want to be traveling to wherever I wanted. Why do I want to go to an office? There's only a few of us, we can all sit and talk over Slack. It was very straightforward. It was irrelevant. It was like, we don't need to be in an office, it's unnecessary.

You reduce the emissions in that way, and then you've reduce emissions in another way, but we'll just use the cloud and then we'll reduced emissions by, well, we'll stick constraints around the amount of code we use. You just start to put those constraints in place and then the snowball effect is that actually we had a tiny, tiny, tiny amount of money that we were spending with AWS, but that we also had a huge amount of people that we could reach. If people stopped using it, the budget, we were spending almost nothing. From a cost perspective, my CEO basically never bothered talking to me about the budget. It just was irrelevant. It was like, "It just costs us what it costs us."

I remember a conversation and he said, "Well, it costs us," I think he said it was $1.00 to acquire people, and he goes, "How much did it cost you to run this platform?" I told him the number and he goes, "Well, that's less than the $1.00 a person then isn't it?" I went, "Yes," and he goes, "Well, that works then doesn't it?" He's sitting there going-

Jeremy: Right. The numbers work out.

Paul: ... "Okay, that's fine." It was that ridiculous. When you started to work it out, it was that ridiculous the way that we were running our company, and it was so positive for the company that he just didn't bother asking us how we were doing it. It was like, "You do your thing, just go on with it, you're fine." I think that's where the conversation gets lost in terms of technologists. When they start talking about data centers, especially when you start talking about containers. You want to sit there and you want to go, "I just couldn't care less." But I think one thing that is worth saying is that being serverless doesn't mean you never use a server. I think it's one of the conversations that a lot of people go, "Well, it means you're always using functions. You're always using the smallest... " It's not that.

It's about using the most appropriate technology, and using it so that it simplifies your application so that if nobody's using it, then you get the scales to zero if possible. But there are times when that's not possible. You have to use something, so you use the most appropriate technology for the right reasons. Sometimes that will mean you need to use something like an EC2 instance. It just might mean that that's what you have to do. It's just that, you just have to understand the constraints, understand why, and then understand that if you do that, you then have other concerns that come along with that.

That's what being serverless means. It's not about no servers, it's about understanding the constraints, understanding the platform, and then building your application. Understanding that you will then need to manage that server in a different way, and understand the application life cycle and retire that server at some point or change it. It's a whole lot more holistic than just, "Oh, we just-

It's a whole lot more holistic than just, "Oh, we just use FaaS." So I think there's a whole conversation there, but it does, for me, come all the way back down to, how can we reduce our impact? Because one of the other things is if you can reduce the impact in terms of operations, and if you can reduce the impact in terms of maintenance. No maintenance over time, significant carbon impact, because you're not having to go back in and change stuff. You're not having to make more code. You're not having to test stuff, all of that. The knock-on impact is huge.

Jeremy: Right now, if I can summarize what you just said, what I got out of that was that Kubernetes is bad for the environment. So I'm just going to use that as the takeaway-

Paul: And go with that.

Jeremy: And go with that. So the other thing that you mentioned though, this idea of working remotely, maybe a benefit of COVID, which I don't think there are any benefits of it, but the experiment of, can most people work from home? Now, I fully understand, now again, get back to the Black Lives Matter and under-represented communities, things like that. They are just not possible for people to work from home. Eventually kids are going to have to go back to school and teachers are going to have to be in the classrooms and you're still going to have to have your janitors, you're going to have your baristas, you're going to have your doctors.

You're going to have that whole range of the economy still needs to exist. But a lot of people in tech are very privileged and the ability to work from home is something that, I mean, I know a lot of people say, Oh, I want the socialization, and that's fine. But think about it this way, I leave my home, I don't turn my thermostat down to zero degrees and let my house freeze when I'm away from home and at the office, I still have to heat my house.

So I'm heating an empty house, even if it's at a lower temperature while my employer is heating a 10,000 square foot office. And so we're heating two places, we're wasting a lot of energy and I spend all that time traveling. I spend, well, not time-travel, but I spend all that time traveling to work, commuting to work.

And then, I wish I could time-travel, that would be nice. It could probably make a few things better, but that's the Back to the Future reference. But if I could, the amount of gas that I put in my car or the train that I ride or whatever it is, those efficiencies there by having to power two places when you don't necessarily need to. And I think what we're going to see, and I'd be curious, your thoughts on this, but I think what we're going to see is companies starting to build much smaller in-person offices, right?

They could be regional, but I mean, having an office building that can fit a thousand employees just might not be necessary anymore. It might be that having an office building or an office space that has room for 50 employees and a few conference rooms and things like that where certain people can go in on certain days.

But for the most part reducing that overall footprint, I think that's going to have a huge impact. And I think companies are going to need to start thinking about how they reduce their footprint because I'm pretty sure that at some point the hammer's going to come down and these companies are going to have to rethink their entire global supply chain.

Paul: Yep. And I agree with that completely. I think we are going to end up and I think tech is probably going to be the lead for an awful lot of this. We are an incredibly privileged group, that we are essentially all working on computers. We all work... Basically need a computer and an internet connection and we can do our jobs. We don't even need to be at home. We can be at a coffee shop, which is not even fair for most people. It's like, well, I am working from the coffee shop all day. Well, stuff you, everybody else.

Jeremy: While you are fully garbed in PPE trying to treat patients in the emergency room, exactly.

Paul: We will happily sit here with our latte and a Danish, it's like, well, thanks.

Jeremy: And we gave you TikToK and Facebook and Twitter, so stop complaining. Yeah, totally agree.

Paul: We should recognize that privilege. And I think we should be at the forefront of trying to make society better. And I think we should understand that. That includes recognizing that moving away from offices is going to have a significant impact on the economies around those offices. So there are going to be shops and coffee shops and things like that, that are going to have to change and grow and move. And so I think we do need to recognize that impact is going to be huge.

But I also think that we need to think about tools and using serverless to build those tools is very good. But I think we need to think about tools around how we make teams better, how we make organizations work better. And it's not so much just the video. The video is lovely, but it's a brief conversation I had with someone yesterday was like, are we doing the storming and the norming and all of those kinds of things anymore?

So, are we actually creating teams? Normally when you get together and you're basically all together in an office, you have the fights and then you get all together, you then get back together and you figure out how to... I'm not sure that you do that over a video call anymore.

So, there's all these other things that we're going to have to start learning how to do differently. And I think technology and technology companies are going to need to be at the forefront of all of that. And so I think we're going to need to learn how to do all of those things a bit better.

But that will have an impact on carbon. And I think we'll be doing things like smart homes better than just having a nest controller or whatever it is in your home.

I think we'll be doing things like repurposing office buildings into something else, possibly homes, possibly something else. Because I think that there are going to be a decent amount of empty buildings and I think our city centers may change. This might take 10 years, but I think we're going to see some real changes in the way that cities work.

But this is all a much bigger, broader conversation. Coming back to the serverless side of things, I think we are going to have to... I think we could well see a change in the way that companies approach technology as well. I think we're going to see that we can't just throw money at tech. I think companies are going to want to see returns. They're going to want to be able to manage and understand how technology is used and where it fits. And that I think serverless is probably better placed than it realizes for that.

Jeremy: Right. Yeah. I totally agree. All right. Well, we've been talking for a very, very long time, and I appreciate the time that you have, if you have a few more minutes, though, I do have some questions for my Serverless Chats Insiders list. And again, if you want to ask questions to guests like Paul, go to serverlesschats.com/insider, sign up for the list, and you can do that.

All right, so first question is from Eduardo. And he asked the question, does serverless contribute to climate change in a good or a bad way? And as cloud providers are investing in more green infrastructures, what does that mean to users?

Paul: So, the difficult thing is that it is probably a net good in terms of serverless, it is probably a net good, because I suspect... It is impossible.

The reason it is hard to answer this question is, and I have asked this multiple times over years now, it's actually very difficult to build an equivalent application, not serverless and serverless, and then work out what the actual electricity usage is. And then, in terms of how do you basically calculate load? How do you work out how much electricity is being used and for how long and over what period, and it's actually incredibly difficult to do.

And also, take the pet shop example, which is the one we all grew up learning, what the heck it was. And, it's like, how do you build pet shop in serverless? It's like how everyone would do it differently. And so what is the reference implementation? There isn't one.

So we don't know, but if you take the idea of function as a service, and if you take the idea of just using... Your functions only do one thing, single responsibility principle. If you take the idea of not running relational databases, but running something like DynamoDB, some completely different idea.

If you take that, I suspect that you are probably net positive. It's a net good. Do your own research, have a look at yourself. But that's what I'm going on. Because I think that the only real good proxy I have is the cost and the majority of serverless implementations are reduced cost. And it's a pretty good proxy for the amount of electricity, somewhere along the line.

Jeremy: Right, you would think that if the cloud provider is charging you less than they're paying less, which means you're probably... And energy and electricity is probably one of those things. I mean, the other thing I would say about that too, is that, obviously, you have a very large server that is running all these multi-tenants, that is running firecracker or whatever it's running under the hood there. The hypervisor in order to spin up these little containers that are your Lambda functions or whatever your FaaS of choice is.

Obviously that server's running, the whole server's running. Can't shut down parts of the server. There's probably some efficiencies obviously to the CPU that if the workload is low, then it's not burning as much electricity, but you still have to have a lot of servers turned on to be able to handle the spike in load.

You can't just have somebody saying, Oh, wait, now we've got every server's used. Let's turn on another one. Capacity planning is still a thing that needs to be done in these data centers. So even as efficient as serverless might be from your implementation, I think at least what you're saying is even though those servers are running full-time, I am only using energy when my part of it runs. So I'm at least reducing it a bit. So I agree with you. It's a hard question to answer, but I-

Paul: It's an impossibly difficult one.

Jeremy: I'd like to feel better-

Paul: What was the second part?

Jeremy: The second part, was there one? Oh, what does it mean for end users? But I think you've sort of answered that.

Paul: I did yeah, I did answer that.

Jeremy: All right. So, Mark asked another question, and this is, again, going back to probably putting AWS in the spotlight, but AWS has the worst credentials for powering their data centers of the big three. And this is what Mark says. They also have the most comprehensive and power efficient serverless offering, presumably even more so once we have Lambda on Graviton2 which lets us use less energy overall to run our apps. So, how should we look at the trade offs? So Amazon, maybe not as green for the data centers, but a much more efficient serverless offering. How do you make that trade-off?

Paul: Yes. And this comes back to the conversation around... It's a difficult one. And again, this trade-off is complicated. I think if it's complex to build an application it's going to be complex to maintain and complex to manage at the end of the day.

So, if you are finding that it's difficult to build, difficult to take forwards, you need a bigger team, whatever it is. I think the amount of emissions you create doing that is probably, in terms of the people, in terms of the amount of maybe get meetings, the conversations that you have around it, the longterm effects are probably going to be higher than if you take the more efficient offering and then choose to do your own offsetting, take an approximation of an offset in some way, shape or form.

I think that's the way I approach it anyway, which is to essentially say, yes, I know they're not perfect because no company ever is, but yes, I know AWS is the worst of the three and I've looked at all the offerings very carefully, but I look at it from the point of view of, but I know how this serverless offering works. I know how efficient it is. I know that it will give me the best offering for my users.

So the likelihood is that I will spend less time in support. I will spend less time trying to work around the things that I know aren't particularly good in all the things that aren't particularly working in the way that I want them to work in other platforms. Not to say that you couldn't work around the constraints and the other ways, but I can see that I know how much less effort it is in AWS than it is in the other two clouds to produce a good application and to manage it and maintain it.

And I've tried to build in all of them. So, I see that efficiency in terms of time as being a carbon impact as well. And that's the trade-off that I see. So, if I didn't see that, I would be moving over, does that make sense? So, that's how I look at that trade-off.

Jeremy: Yeah. Makes sense to me. All right. So, one more question. This one is anonymous and I'm assuming, because it's such an easy question that they didn't want to be named when they ask us an easy question. It says Paul's articles on serverless have been thought provoking in a good way about how he sees and has seen service evolving with the industry adopting serverless more and more. I wonder what he sees in the future of serverless and what he sees as the answer to what is serverless 2025? So simple question, see if you can...

Paul: Thank you very much. Where do I see serverless evolving and what do I see serverless in 2025. Well, thanks for that, whoever sent that in. Oh, that's so, okay. I see tools, I don't think we have the tool sets yet to properly build serverless applications. I see we have an awful lot of people building web applications on top of serverless. I don't see we have a lot of applications building the ability to simply connect.

So at the moment, I think we have things like SAM and CloudFormation and all those bits that allow us to automate, but they're complex and they're not straightforward. And then we've got other tools on top of those, which your serverless frameworks and your others, I can't think of the top of my head.

And you've got all these, but I think they are not yet evolved to where we need everybody else to get to. And I think until we have that next evolution of tools, and I think I haven't seen it yet. I haven't seen that tool yet that allows someone to come along and go, I just want to do X. And that it's like three or four lines of code. And I think Begin and I think Architects and all of that kind of stuff and all of that, I think has some of those elements, but it's not quite there yet. I think there's a little bit missing a few pieces, but I think it for what it does, it's brilliant. So I think we're just missing some of those tools.

I don't think we'll see a multi cloud tool. I think the main three clouds in terms of serverless are so different in the way that they've approached function as a service, data and all of those things. I think we will see that they will diverge. I don't think compute is the key anymore.

I think we will also see serverless as being more about data. I think we will see it less about compute, and more about data. And I think we will see the divergence in the clouds being about how people store data, how people use data, how people build pipelines around their data and how people move data at speed, get queries out to their data. I think that's where the innovations will come. So basically building the building blocks, I think we'll get more of that. And then the keys around the data.

So, we will see less relational databases. Yes, I really can't stand them. I think we will see some really, really fascinating data tools appearing. That's where I think 2025 is going to be. So when someone asks what serverless is going to look like, I think we probably need to be looking at the... I think what people are going to be really scared of, is it's going to move away from that three layer, which we still think about the three layers, your client, your server and your data and all that kind of stuff. We still think in that world.

I think it's going to be blown away. The data layer is going to all of a sudden learn to compute and it's going to learn to do all of these clever things that we didn't realise it could do.

And then all of a sudden these data scientists going to walk in and essentially become very, very much more key for everything we build and that's where serverless is going. So when we get to that level, when we start to go down that road, and I think some people are there, but not everybody. I think that's when we start to see what, away from microservices, that I think is not where serverless is going, that I think is where we're heading. So anyway, that's my broad thinking on where 2025 is going to be. How's that?

Jeremy: Yeah, no, I think that's great. And I agree, I think that one of the most difficult things about serverless right now, despite the lack of tooling in many cases, is the fact that everything is a primitive, right? And you've got a Lambda function that needs to connect to SQS and it needs to connect to Dynamo or whatever. You have this whole, all these little tiny primitives.

And what we really need is, the building block needs to be more abstract than just those tiny things. And as those patterns emerge and we are already seeing a ton of them, and as we start to encapsulate those into things like the CDK and into serverless components is something that is an interesting thing that they're doing. Like you said, Architect, and some of these things that abstract away some of the more complex individual building blocks and make them much larger pieces that you can put together, that's interesting.

And then the data stuff you're already seeing this, you're seeing Adobe, you're seeing Salesforce and some of these other ones, I think Twilio is doing it now too, where you can run serverless functions on their platform in reaction to data or in reaction to the events that data changes in your system.

And that's going to be really interesting because then that's hands off the compute and then the data handling piece of it gets smarter. But, I mean, anything else? I mean, we've covered quite a bit here, but any last words on serverless and going green?

Paul: No, just more serverless and go green. I think it's just very much a case of get on with it. Become an activist. I think it's what we all need to be doing.

Jeremy: Awesome. All right. Well, Paul, thank you so much for this conversation. This was excellent. I had a great time. I hope you enjoyed it. Hope the listeners enjoyed this. I think this is a little different than what we normally do, but honestly, this is stuff that I'm passionate about. I know you're passionate about it. So anyways, if people want to learn more about what you're doing and learn more about green energy and how they can make an impact, what's the best way for them to contact you or find out more?

Paul: Just Twitter. I'm on Twitter @pauldjohnston. And just also find me on roundaboutlabs.com. Just my details are on there as well. So, either way is absolutely fine.

Jeremy: Awesome. And then your blog is medium.com/pauldjohnston as well.

Paul: Yeah, that's correct.

Jeremy: Awesome. Thank you again. I will get all this information to the show notes. So if people want to go check out the show notes, they can find the information there. Thanks again, Paul.

Paul: Thank you for having me.

This episode is sponsored by New Relic and Homeschool

Episode source