🎂 Switch to Cloudflare to cut carbon emissions
Presented by: Annika Garbers, Patrick Day, Andrew Daly
Originally aired on September 27, 2023 @ 12:00 AM - 12:30 AM EDT
Welcome to Cloudflare Birthday Week 2023!
2023 marks Cloudflare’s 13th birthday! Each day this week we will announce new products and host fascinating discussions with guests including product experts, customers, and industry peers.
Tune in all week for more news, announcements, and thought-provoking discussions!
Read the blog post:
- Switching to Cloudflare can cut your network carbon emissions up to 96% (and we're joining the SBTi)
Visit the Birthday Week Hub for every announcement and CFTV episode — check back all week for more!
English
Birthday Week
Transcript (Beta)
Hi, everyone. Welcome to Cloudflare TV. Welcome to Birthday Week. My name is Patrick Day.
I'm head of impact at Cloudflare. I am very excited to be here today to talk to you on our first day of Birthday Week this year about sustainability.
We're publishing a report versus the analysis. Mason is publishing a report this week that we were proud to work with him on.
I'm really excited to be here with one of the authors of that report, Andrew from Analysis Mason, and then one of our heads of sustainability at Cloudflare, Annika as well.
I'll let you two introduce yourselves, but yeah, really excited to be kicking off this week.
It's one of the most exciting weeks of the year here at Cloudflare, as you may know.
So yeah, excited to get started.
Annika, do you want to go first? Sure. I am so excited to be here too.
Birthday Week is the best week of the year. Yeah, I'm Annika.
I'm on the product team. I'm responsible for Cloudflare's network services product portfolio, which is all about helping our customers transition from legacy network architecture, I think stacks of hardware that used to live at data centers and did things to help secure and connect our customers' networks and moving those, consolidating them to cloud -based services, which turns out also has sustainability benefits, which we'll talk a lot more about today.
Andrew, you want to go next?
Hi, everybody. I'm Andrew Daly. I'm a principal at Analysis Mason. I'm actually based in the UK.
I'm very much looking forward to talking to you a bit more in a few moments about the work we've been doing with Cloudflare, a really interesting project and a really interesting result actually as well.
So I think we started this process, Andrew, almost a year ago at this point.
So it's actually a really exciting day to sort of get this report out, be able to talk about it publicly.
I know all of our engineers and folks who've reviewed it over the last year or so have been really excited to start to talk and share the information with people.
So why don't we just jump right into that? So I'll finish the introductions quickly.
I'm Patrick Day. As I mentioned, I'm head of Cloudflare Impact based in Washington, DC.
My team handles Cloudflare sort of impact programs, which help support journalists and human rights organizations around the world, protect elections, but we also do our sustainability reporting and our ESG reporting, our carbon accounting, and we also do some human rights work as well.
So I think it's something, this is a project that Anika and I have been passionate about at least for several years now, working with AnalysisMason for the last year.
But Andrew, if you want to talk about sort of the report that AnalysisMason is publishing today and sort of walk us through the key findings and how you approached it and we'd love to hear about it.
Of course. Let me say a few words about who we are for those of you that don't know us.
So we're an independent management consultancy.
We focus on the TMT sector, so telecoms, media and technology.
We've got about 17 offices around the world and about 450 staff.
And this is actually our second piece of work with Cloudflare. We previously worked with you on a piece on caching, but this piece of analysis, this report, so Cloudflare commissioned us to calculate whether moving enterprise network functions from on-premises equipment into the cloud could actually create carbon savings.
And I think that is a concept which is intuitive, that these things should create some carbon emissions savings.
But I think what we did on the study with a lot input from Cloudflare is actually put some numbers to this.
And as I'll explain in a moment, we got quite an interesting and exciting result.
So what did we do?
Well, we essentially modelled networks. That's what we do at AnalysisMason.
We modelled enterprise networks. So the network connectivity needs of businesses of different sizes, and we modelled them under two scenarios.
So one is where the network functionality is provided by on-premises equipment, so either in the corner of an office or maybe in a data centre that the business owns.
And in the other scenario, we modelled the same demands, but where those demands are met by Cloudflare's products.
And then we calculated the energy use of the two and the associated carbon emissions, and we compared the two results.
It's probably just worth saying before I go on to the result that we defined quite a specific range of functions that are in the scope of the analysis.
So things like routers and switches are likely to have to stay on -premises, so they're not in scope.
But lots of other functions around network management, network optimisation, network security can be moved to the cloud quite effectively.
So those are the functions in scope.
I'll leave it to the report to list those out. I won't attempt to do that here on the call.
So what did we find?
Well, we found that when you compare the two on-premises to cloud-based, it shows somewhere between 80 % and 90% reduction in the carbon emissions that are associated with generating electricity for those functions.
So again, I think that's perhaps the result we were hoping for, but it was great to actually put those numbers on the problem.
So I think that, just to jump in for a second, I think that's the part that is most exciting, not only for Cloudflare, but sort of in the context of Birthday Week.
So Birthday Week is all about things that Cloudflare has done to give back to the Internet or to help build a better Internet for everyone.
And so typically we've launched new products or new services or made things available for free to everyone.
I think this is one of the first times we've really talked about Cloudflare's sustainability impact or potential impact on the Internet generally.
So my team, we do a lot of internal carbon accounting, so figuring out where our data centers are and how much energy they use and publishing that information.
But this is really the first time we've had a chance to step back and see what Cloudflare's...
That's all about our impact internally, but it's exciting to hear about applying this technology externally and what it means for the Internet as a whole.
So that, I think, is one of the reasons we thought it was important to put it into Birthday Week, because it's something that Cloudflare contribute to the Internet ecosystem generally, which is very much what we've tried to do in these weeks.
So sorry for interrupting you there.
That's the part that just makes me sort of feel proud and excited to start to share this information.
Yeah, definitely. I mean, it's worth saying a few words about where that gain comes from.
So as we show in the report, most of that gain comes from the fact that cloud infrastructure is so much better utilized than on-premises infrastructure.
So you can imagine a business, they might have a firewall in the corner of their office.
That firewall is on all the time and it's being used some of the time, some of the days of the week, some of the hours of the day.
It's also probably much bigger than it needs to be because the business has to over -provision to cope with peaks.
When we look at cloud infrastructure and Cloudflare's platform, that's aggregating demand from hundreds, thousands of organizations.
And it means the utilization of the infrastructure of the power use is so much higher.
So that accounts for most of our gains.
We also looked at things like the efficiency of data centers where cloud infrastructure tends to be located.
That tends to be much better than on-premises.
And we also found some gains related to the local electricity grid.
But again, I think what was really interesting on the project is to actually separate out those different gains and see where the savings come from.
So one of the things I was curious about, I think there's a couple of well -documented studies that talk about moving applications or different services to the cloud that generally yields the type of savings that you're talking about, particularly in utilization gains.
I was curious when you did your literature review, if there was another study that also looked at network enterprise functions.
I was not aware of one. I just didn't know if you all found anything. No, there aren't many that we're aware of.
There are other studies that, as you come at the utilization gain idea from different directions, some are more sophisticated than others.
I think that this study is the only one we're aware of that's focused on that specific use case that enterprises can actually think about.
I think that's another thing that's really exciting.
Arunka, sorry, I saw you starting to say something.
No, I think if you look at the industry trend, it kind of makes sense that that's where the focus has historically been, because really networking and security functions have lagged 10 to 20 years behind compute and storage in the transition to the cloud in general.
When customers come to us to ask for help solving problems that are related to connecting, securing their networks in this new cloud-centric world that they're having to deal with, where their applications have moved to the cloud and their users can now work from anywhere, but they're still managing maybe this legacy stack of networking and security gear, it's clear that this component of the industry, the networking and security, just hasn't really kept up.
I think that you can see that in the literature that's available too.
We haven't even really started the conversation or this study is the first of its kind in the conversation around how transitioning the network and security functions can deliver something.
I'm sorry. On the one hand, I thought it was really encouraging that it was sort of consistent with other studies that had talked about moving different types of applications to the cloud, but on the birthday week exciting part, it really was the first to talk about that functionality specifically or those services available on the layers of the Internet stack that Cloudflare operates.
Actually, I don't mean to sort of steal the conversation, but it was almost sort of a natural segue.
Anika, you sort of described what your role in talking with customers and helping with their migration and dealing with legacy systems and moving to Cloudflare.
You touched on this a little bit, so it may be a tad repetitive, but give us sort of just a 101 of your typical conversation with customers, why they come to Cloudflare, what types of services they're using, what they're hoping to get out of our platform that's different from what they've had previously.
Sure. Cloudflare started as an organization over 10 years ago, I guess 13 years ago, it's birthday week, focused on our customers' web applications and then APIs.
Anything on the web, we were trying to make it more secure, faster, and more reliable because what we realized is that customers didn't have great solutions to solve those problems that didn't rely on either legacy hardware or these really expensive services that were geared at kind of just the world's largest enterprises.
Everyone that wasn't a company like that didn't have a good solution.
We started with a focus on websites and HTTP applications. Then over time, our customers have asked us questions about, okay, well, I've got all this other stuff connected to the Internet too.
Could Cloudflare sit in front of an entire data center?
Could Cloudflare help secure my private network infrastructure?
The reason they want us to do that is because they're facing these challenges that I mentioned before, those kind of two shifts that have really fundamentally changed the way that corporate network and security works, which were applications moving from the data center to the cloud, and then users being able to largely work from anywhere.
The traditional model where you send all of your traffic through a centralized data center in a stack of hardware functions like firewalls, intrusion detection systems, VPN gateways, the types of hardware that we actually analyzed in the migration with this study, that model doesn't work.
Instead, what Cloudflare does is we deliver all those same security and connectivity services with one single control plane across them, but from our distributed global network instead.
Wherever you are in the world, I happen to be in Denver today, I'm connecting to the Cloudflare Colo here in Denver, which is where my traffic is getting filtered for security.
When I go home to Atlanta, I'll connect to the Atlanta Colo and I'll have those controls there instead.
But all of the servers across our entire network run that entire stack, and then also for all of our customers, which helps deliver those really high utilization per server that gets us some of those gains in the power efficiency.
We're doing more with every watt of electricity we're consuming.
I know you were in Australia for Cloudflare Connect.
So Australia, Japan, Denver, Atlanta, wherever you go. You got customers all over the world.
That's great. Andrew, one thing I wanted to ask you, which I thought was particularly interesting, and it's something that Anika just touched on, was as we were going through this process, we started last October really in-depth and going through and starting to prepare, and then we got some initial results in December on just the WAF service that we published.
It was around the same ballparks for that 90% potential carbon emission savings.
That was exciting in and of itself, that just comparing one service in isolation to its on-premises counterpart, based on the utilization gains that you were talking about, really was a fundamental shift in terms of carbon footprint.
Now that we're at this final stage of the report, and you mentioned there's almost too many products to list here that were considered and their on-premises counterparts, talk a little bit about what I really found interesting with the utilization gain.
It's not just that moving one service to the cloud is more efficient.
In Cloudflare's case, it's because we've replaced so many different types of functionalities that used to be served by different types of on-premises hardware, and the utilization gains are sort of spread across those services.
Was that sort of, that was the most interesting part of the study to you.
I'm curious as you were sort of working through that, if there's anything that stood out to you about that or surprised you in any way.
Yeah, the whole thing was very interesting, and seeing how different functions consume power in different ways was quite interesting as well.
On a related, sort of related to the points you just raised, Patrick, it's quite interesting to see how the different sizes of business have slightly different results.
So if you are a very small business, then there's only so, it's only so small you can go in terms of your on -premises equipment, and so the gains you get as an even smaller business tend to be a bit higher.
If you're a large business, then the equipment tends to be bigger and, you know, better suited to handling larger amounts of traffic, and so the gains are a bit different.
But actually seeing how those sort of power-to-traffic-demand relationships play out was quite interesting in the study, and talked to that point you were just raising.
I'm sure challenging a little bit too to try and parse out all that.
I know there was a considerable amount of work done on that piece.
One of the things I thought was really refreshing was that, you know, we're familiar with sort of other studies that have documented shift to the cloud, and that there's potential gains to be made based on utilization or PUE or whatever, and in this case, you know, one of the things I was most interested about was, I sort of lost my train of thought, actually.
I guess I'll just share, I had an important question there, but it's now escaped my mind because there's too long of a show, but so Andrew, I'll just ask you sort of, you know, broad strokes, conclusions, you know, lessons learned, things that could lead to future work, things analysis makes us interested in, sort of along this same transition.
Yeah, of course.
I mean, why don't I say a few words about the process, because that was quite interesting as well.
So on the on -premises side, as you'll see in the report, we did a very extensive benchmark of equipment performance, and that, most of it was publicly available, it had its own challenges, there were lots of vendors, lots of different models of equipment that we needed to be reconciling.
On the cloud side, we had to come up with quite a novel approach, actually, to allocate this shared resource between the different services using the shared resource, and we worked with Aneka and some of the other members of the Cloudflare team to think about the relative computational load of different functions, and that took a bit of time and thinking to get right, but I think we got something that's quite robust, actually, in the end.
Just to your point about next steps, I think we're all quite happy with the study.
I think it would be interesting to revisit it in time, you know, the efficiency of cloud servers will continue to improve, the way they pull away from on -premises is going to change, that would be quite interesting.
We'd be very interested in doing similar studies on other functions and products that cloud providers like Cloudflare provide, so I think those probably two things are areas on our mind at the moment.
Aneka, can I ask you a question with sort of your product hat on, because that piece that Andrew was talking about, and sort of when you're designing or working with products in the engineering team, sort of the idea that you're sharing compute resources and that whatever you're developing sort of takes a piece of that, which ultimately turns into energy and emissions for this conversation, but I'm curious if that comes up in sort of your day-to-day when you're sort of looking for product efficiency or, you know, trying to make the most use of every watt of energy on a Cloudflare metal, like how that sort of, how you encounter that?
Yeah, absolutely. I think one of the challenges that we anticipated or realized that we would have early on when we started looking at this was, okay, the study kind of quantifies, looks at the individual hardware components that are delivering these functions in a legacy stack, right, like let's look at the power footprint of an average next generation firewall or the power footprint of an SD-WAN device, but as we mentioned before, all of Cloudflare servers are a general purpose and run all of these functions in one place on one server, and then also for all of our customers, and so how are we going to kind of piece that out and determine what percentage of the power consumption of a single server is associated with each one of those functions?
And so we looked at a couple of different ways to kind of gather that data, but one of the things we realized was available to us was actually a resource breakdown per service in terms of CPU and memory consumption on our servers, and the reason that we keep track of this information is to be able to understand what services are using a lot of resources and then figure out where we need to focus to drive efficiency, and our teams have been doing that for years for reasons other than reducing power consumption for sustainability efficiencies, right?
It's like, yeah, we want to be able to do more with less to deliver cost benefits to our end customers, and so we've got this kind of resource inventory or leaderboard or loserboard, however you want to think about it, of which services are consuming the most resources on our edge to help us understand and also do things like capacity planning, right?
If this service that we've introduced that's new is growing a lot over time, how do we expect that to influence the additional capacity that we might need to provision at our edge so that data is available to us for those other reasons?
And I think this is just an example of, there's many of these that we've sort of seen in other conversations around sustainability impact at Cloudflare, but it's one of those examples where making the right decision for the product and for a cost efficiency perspective for our customers and their experience actually also has a corresponding sustainability impact, which is really cool.
I think that's one of the best, I think that's why potentially in the last three to five years you've seen such sort of embracing sustainability across sectors and companies of all sizes is that there really is often a cost savings associated with it.
So it's not only good policy, it's good for the planet, but from a business case, there generally are, in my experience, cost savings to be found as well, which helps everybody.
I'm wondering, Anika, in those conversations you talked about with customers, if they're handling migration, has sustainability come up in those conversations?
Is it sort of increasing over time? Has it always been there? I'm curious what you sort of experienced from them.
Yeah, for sure. You know, some people characterize product management as like, you're the people that come up with the ideas for what to build, but I've never had an original idea in my life about the product.
It all comes from our customers. I'm not sure that's related to that.
No, but that's real, right? You hear things, being out in the world, talking to customers about the problems that they have, and I think especially our customers in European countries who are looking ahead to what the global regulation landscape looks like and reporting requirements coming out around sustainability, trying to do things like science-based targets, initiative commitments, where they're going to drive down their carbon footprint over time, they're looking to their vendors, everyone in their extended supply chain, and that includes the IT stack, to understand, hey, how can you help me with this?
What can you do to make this story easier?
And it's not just us being transparent and providing information about our own carbon accounting practices and how we're trying to reduce the energy footprint of our services that we deliver, but also as a partner to them, our customers are looking for, what can you actually do to help me improve over time?
Not just understand, but also improve. And so it comes up all the time, and that's why we initially started working on these projects as our customers are asking for it, just like they're asking for new features and all kinds of other things that we'll be announcing this week for birthday week.
Okay.
So I'm going to try not to go into a long diatribe here about climate and regulation, et cetera.
It's sort of my natural inclination, but I'm going to keep it short.
But since you mentioned it, I think the thing I'm most excited about the report is, in the context of birthday week, this is something that Klappler is helping do to build a better Internet.
So that's the first thing, sort of a fundamentally more efficient way to deliver these services, which not only provides sort of cost and business benefit, but also sustainability as well.
The second piece is, what it could mean for our customers.
And you mentioned it. And so this, okay, I'll try to keep this short.
So very broad strokes, and Anika has heard me, I think do this before, but if you started sort of the international level Paris climate agreement, there's a global consensus that we need to limit sort of global emission gain to 1.5 degrees above pre-industrial levels by 2050 to avoid sort of the worst effects of climate change.
So then you go down a level to sort of the national or regional level.
So you're starting to see more legislation like the European climate law or the European green deal to down another level where you're starting to see sort of regulatory action by the U.S.
Security and Exchange Commission, which has published a draft rule that will require companies to publish not only their direct emissions and the emissions they generate from energy consumption, which is what we talked about mostly in the context of this report, but also supply chain.
And that's a really big one. And I'd say sort of an increasing part of my day over the last year or so has been talking with customers who have either set carbon reduction targets, which Anika mentioned through SBTI or some other initiative.
And they're really starting to go through their supply chain vendor by vendor, try to get whatever climate or emissions information is available, and then working together to sort of create a better climate outcome.
And those are all being driven sort of from Paris, the climate agreement down to the national level, down to the regulatory level.
And that's, and I should say a big part of this is also consumer interest, right?
If you all the way down to a consumer product, consumers just sort of, the data is pretty clear, they want more sustainable solutions.
So you have all these pressures. And I think you're starting to see rapid adoption by not only fortune 500 companies.
I think the article I looked at the other day was 63% had set some sort of emissions reduction target by 2050 consistent with the Paris agreement, but it's increasing every day.
And the numbers are sort of staggering.
So I'm hearing our sustainability team, which typically has been working on initiatives at Cloudflare and working on calculating our footprint and getting that data out.
It started with sort of investor interest, like it's becoming more of a gauge of future company risk investors, community wants to know sort of what your emissions footprint is.
But it's now because of all those sort of pressures and new commitments, particularly on supply chain, I think that is sort of the most interesting area of the moment.
You're starting to see customers seek out directly sustainability teams at every company within their supply chain to find out what their footprint is, how can you be an asset in meeting the climate reduction targets that I've set for my business.
So that's really how this has evolved. And as more and more companies set those supply chain requirements, it sort of has a cascading effect.
So that is a very long winded, I could have said it shorter, but that was partially long winded.
It's a very long way to say like, I think the second most exciting part of this report is we can help our customers meet their own climate goals, right?
If you're still running sort of these type of on-premise equipment, it's really worth taking a look at what that footprint is, what types of savings you could achieve by moving to a platform like Cloudflare, and we can sort of help with that.
And so that's, you know, it's two things.
It's giving back to the Internet, you know, there is a fundamentally more efficient way to organize Internet architecture.
And that's sort of been Cloudflare story on the business side, the security side, and now sort of increasingly on the sustainability side, which is really exciting.
But then also sort of being a good news story for our customers who really are looking for these emission savings.
They, I think, you know, I don't know if we're at the inflection point now, but over time, I think you'll start to see migration towards more efficient, more climate friendly products generally, as we get sort of closer to those larger commitment dates.
So those two things, I think it's really, we've had sort of good news to share in terms of our sustainability commitments that we've made as a company.
So Cloudflare is powered by 100% renewable energy.
We're working to mitigate our historic emissions footprint by 2025.
So we felt like we've sort of taken responsible steps to account for our own footprint.
It's really nice to be sort of in a receptive position and see, you know, organizing through a more efficient architecture like Cloudflare, it's good for the Internet, can be good for our customers.
And so it's sort of a good news story all around.
And I'm excited that we are, it's been a long process. I know Anika has been in many meetings, Andrew and I have seen each other almost on a weekly basis for a year now.
So it's exciting to be sort of near the finish line.
Anything else? I think we've sort of covered it. I think the report and Anika and I put together a blog, which talks about in more detail, which is available on Cloudflare's website, and all analysis.
Mason will have access to the report on their website.
But anything else? I know I talked for a long time, questions about, I feel like I've asked you a lot of questions.
So it's only fair game if you have sort of questions on the legal regulatory side.
But other than that, concluding thoughts before we wrap up today.
Nothing major for me. As I hope you said earlier, it's been a really interesting project from our side.
And, you know, it's not just me, there's been a whole team working at Analysis Mason from quite a broad range of sort of backgrounds and areas of expertise.
But yeah, it's great to see us get to this point.
And yeah, looking forward to the report launch.
I'm glad you said that. Yes. The coolest thing for me has been seeing how holistically Cloudflare is thinking about this space.
So today we talked about, you know, primarily the benefits that our customers can see by consolidating these functions to the cloud, and then a little bit on the energy and carbon emission side.
But there's also lots of work going on within our infrastructure team to improve the efficiency of the new hardware that we're deploying in our service, our data centers, and do things like component modularity to make sure that we're thinking about end of life for those devices responsibly.
There are tools that our customers can use to make sure that they can operate their services that they're using with Cloudflare from locations that use underlying renewable energy infrastructure within the Cloudflare dashboard.
There's reporting tools that we make available to customers in the dash for them to understand their impact of using Cloudflare services.
And that's just a tiny little sliver. There's tons and tons of information if you go on the Cloudflare blog or search Cloudflare TV for sustainability.
Lots more content. So if you like this topic and you want to dig in more, there's a lot out there.
So the last thing I'll say is, and I'm glad Andrew mentioned it, it was wonderful working with the team at AnalysisMason.
There was quite a few analysts and technical experts that we had the privilege to work with over the last 12 months.
And I'll also say a big thank you to Annika in place of the dozens of product and engineering and infrastructure folks who took time to meet with.
Sustainability is something that's important to us at an employee level, as a company, and then it was just a pleasure sort of working with the AnalysisMason team again for the second time.
So I think that's it for us today.
We've got a lot of birthday week left after today. So please stay tuned.
There'll be many more announcements, but Annika and Andrew, thanks for your time.
And we'll talk to you soon. Always a pleasure.