Johannes Lintzen 0:00 Not everyone implements the TLS standard as they shoot. And from our early investigations, we saw that some middle boxes, some software, if they see these bigger keys, even though they should just go negotiate old stuff, instead they break.
Bas Westerbaan 0:16 Howdy, folks. And a big warm welcome to Shielded, the last line of cyber defense, the podcast brought to you by the fine folks at PQShield. We're not just yacking about why post quantum cryptography is important anymore. We're getting down to the nitty gritty how. It's going to be great. I'm Joe Linsen. Welcome to this episode of Shielded, the last line of cyber defense. We all know how hard it is to take a first step on any journey. I'm excited to talk to Bas Westerbahn from CloudFlare today about his experiences in taking not only the first steps, but going the full length. Today, we're joined by Bas Westerban, research engineer at Cloudflare, to discuss the realities of quantum migration. From key management challenges to hybrid cryptography and compliance, Bas shares insights from Cloudflare's journey toward post quantum security and what organizations can and must do now to stay ahead. Let's dive in. So first of all, welcome, Bas. Thank you. Maybe you wanna say a few words about what it means to be a research engineer in this day and age and what your daily job kinda looks like and help us understand, uh, the angle that you come in at.
Johannes Lintzen 1:30 Oh, thank you, Joe. So at CloudFlare, we have a research team where we really try to mix research with we will replicate with what we need in practice. So we have multiple branches, but post quantum is something that for us started back in, I think, 2017 even. So that's when it first started within the research team. And what we really do there is not just trying to do a proof of concepts, but also seeing how will it really work at scale.
Bas Westerbaan 1:58 Well, that's great. So I'm looking forward to finding out more about this in the next twenty to thirty minutes. And then I've heard, Bas, as we've been speaking in the preparation for this podcast that you try to go on a ski trip recently to Norway, if I'm not mistaken, and things necessarily didn't go as planned. Maybe you missed the ferry. Maybe you can share some of that story with our listeners as well.
Johannes Lintzen 2:21 Yeah. There was, uh, quite a different ski trip than I expected. I live in The Netherlands. We wanted to go to Norway and take the ferry from Denmark. But on our way, we had some debris on the Autobahn. Flat tire turned out the rim damaged.
Bas Westerbaan 2:36 It wasn't your tire, though, was it?
Johannes Lintzen 2:37 No. It was my brother's car.
Bas Westerbaan 2:39 Oh, okay.
Johannes Lintzen 2:42 But that's a day full of waiting in the call for the tow truck, make arguments with insurance companies on the phone. It's something, uh, fun and, uh, to look back to.
Bas Westerbaan 2:53 A little bit of unplanned firefighting activity. So I guess the main question is, was the trip worth it?
Johannes Lintzen But that's a day full of waiting in the call for the tow truck, make arguments with insurance companies on the phone. It's something, uh, fun and, uh, to look back to.
Bas Westerbaan 2:53 A little bit of unplanned firefighting activity. So I guess the main question is, was the trip worth it?
Johannes Lintzen 2:59 Well, in the end, we decided not to try to take the boat again, but instead take the long route, get a new rental car, take the long route via Denmark, the bridge, the what's it name? Oresund Bridge, I think?
Bas Westerbaan 3:10 Oresund Bridge. Yeah. Yeah. To Sweden and then Norway, and then
Johannes Lintzen 3:13 actually the longest tunnel, the Leerdal Tunnel, the longest car tunnel in the world. So in the end, we made it a fun detour.
Bas Westerbaan 3:19 Bit little unplanned detour. That's great. And well, speaking of detours, I suppose your background really is in mathematics, and that includes a PhD from Radboud University. And you've worked from anywhere between academia and industry, UCL, PQShield, and now CloudFlare. I'd be curious, and I'm sure our listeners as well, what drew you to sort of the field of cryptography, and how has that academic foundation maybe influenced your approach? You used the term earlier, real world applications of cryptography and now post quantum cryptography. Would you say that background has changed your view on things?
Johannes Lintzen 3:54 So I studied mathematics, and my PhD was on very theoretical side of quantum computing. So not even the more practical bits of quantum computing I care about now during my day job. But the background is helpful, but very much a theoretical side of things. But I did my PhD. My supervisor was the head of the security group. And in that security group, we had, uh, few people working with Post Quantum. Uh, Peter Schwabe from Kiber, Dalesium, I think also an advisor on Tissue. Johan Dammer from AES. And so when I would be procrastinating on my thesis research, I would be having fun with the learning about cryptography, writing maybe accidentally a paper with the folk there. And in a sense, it's not that strange because especially cryptography engineering, it's really at the intersection of mathematics and programming. And I also had a keen interest for that. Yeah.
Bas Westerbaan 4:45 So there's a theme of detours here. I can sense that.
Johannes Lintzen 4:48 Yeah. And going forward, seeing all the work that maybe jumping the gun in. But with all the work we're going to have ahead of ourselves, I think for a lot of people, there will be more linear careers going .
Bas Westerbaan 4:45 So there's a theme of detours here. I can sense that.
Johannes Lintzen 4:48 Yeah. And going forward, seeing all the work that maybe jumping the gun in. But with all the work we're going to have ahead of ourselves, I think for a lot of people, there will be more linear careers going
Bas Westerbaan 4:58 Absolutely. I couldn't agree more. And I think Cloudflare certainly is leading the charge when it comes to the migrations. I'm actually just back from a conference with payments and identity professionals. And again, I was reminded that many organizations are really only just waking up to what the urgency is to take a close look at post quantum cryptography and what it needs to be done within those organizations. I think you can really speak from a background of experience because you've already several years into the migration behind you. So I guess based on that experience, where do you start? Like, what is a good step? What are the first few good steps that a company should and could take when they start that journey and start preparing for PQC?
Johannes Lintzen 5:42 So if you wake up, that's already a very important percent. No. That I mean, once you have executive level buy in that this is a thing that we should care about, that's already a huge deal of the work. So that's really an important first step. So from my work, I really think about this from a technical angle usually, but this is not as much as a technical problem, but it's more a process problem. It's a change management problem as well. Sorry. Life cycle management problem as well. That's, I think, a good way to look at it. So there's a lot to say here, but maybe to speak on some personal experience, we are in the first step, we are upgrading to post quantum encryption. What do you do? You have to figure out what do we need to upgrade. Of course, no one at the company knows everything that's running at the company. Right? You have no idea. So the first step is you ask each engineering team to try to figure out what software are you running, what connection are you running, what is urgent, what is less urgent, what are the steps required. Of course, this requires preparation. Right? You need to know. You need to get an injection. And a part of that is also you need to have the expertise on board to help coach this because, for instance, one of the biggest this is not necessarily deep post quantum crypto expertise, but also stepping back. One of the questions we had in a checklist for each of our teams to answer is, are you using TLS for your communication? Right? Because that means that things are relatively simple. One of the most common questions we got in return not is how do I know if I'm using TLS? Right?
Bas Westerbaan 7:17 Because it's so hidden. e questions we had in a checklist for each of our teams to answer is, are you using TLS for your communication? Right? Because that means that things are relatively simple. One of the most common questions we got in return not is how do I know if I'm using TLS? Right? Because it's so hidden.
Johannes Lintzen 7:18 Yeah. Often, the engineer doesn't directly see it, right, where it is. And also, not every engineer knows what TLS is. So it's part figuring out what is there. It's also having the expertise on board to help coach this. And then for us, it's mostly we write most of the stack ourselves, software stack. But for an average organization, they rely mostly on vendors. Right?
Bas Westerbaan 7:40 Right. It becomes more of a supply chain challenge than something that you can influence directly. I think what you said about getting executive buy in really resonated. So I think that's a really good thing to remember. And then I suppose I've heard other organizations talk about having a center of excellence established, like, a working group within the organization. So it sounds like the group that was asking the questions within CloudFlare was such a center of excellence that took on that task and sort of ran with it within the organization. And then I suppose, I mean, you mentioned TLS as sort of being the first identified use case within the CloudFlare environment. And then sort of going down that route, you've decided to roll out hybrid key establishment, but not, like, the full post quantum authentication, I understand, is still a work in progress. So is there a specific reasoning behind this phased approach? Is that maybe something that came out of those questions that you asked the engineering teams?
Johannes Lintzen 8:37 Yes. Well, this was a very deliberate choice. So I think if there's one takeaway from someone just learning about the post quantum migration you should take, I think it's that it depends a bit on your exact needs. But for most companies, I think there will be not one migration, but two migrations. And the reason is is that there are just different types of cryptography which are vulnerable in different ways. So if you think about cryptography, it's about encrypting so that someone can't read something that they shouldn't. And to achieve that, we use one thing we use critically is encryption, which needs to be post quantum. And the issue is that an attacker, they can store transmitted encrypted communication now. And if it doesn't use post quantum cryptography, in ten, twenty years, whenever they get hold of a quantum computer, they can break that retroactively.
Bas Westerbaan 9:29 That's the threat that is commonly referred to as a harvest now, decrypt later.
Johannes Lintzen 9:33 Exactly. Yeah. Harvest now, decrypt later. I stored out the clip later. Some other people say, post quantum, quantum, and recent quantum.
Bas Westerbaan 9:39 Same idea. Right? 00:09:29] Bas Westerbaan: That's the threat that is commonly referred to as a harvest now, decrypt later.
Johannes Lintzen 9:33 Exactly. Yeah. Harvest now, decrypt later. I stored out the clip later. Some other people say, post quantum, quantum, and recent quantum.
Bas Westerbaan 9:39 Same idea. Right?
Johannes Lintzen 9:41 Yeah. So that is urgent. Right? Because in a way, it feels we're already kind of late. I mean, we serve data for I mean, not all customers care that much about the data in the long term, but some customers do, and we certainly do. Or, um, it is not just that is also a log as passwords. We care about these things also long term.
Bas Westerbaan 10:01 Right. Everything long lived.
Johannes Lintzen 10:02 So that's the first thing we tackle. And it's also relatively easy in the sense that even though the post quantum cryptography, it's a bit different than the cryptography we have today. It's fast, but it's bigger on the wire, but it's still, in the end, performs well enough that it's not that annoying. It's not problematic to deploy. Yeah.
Bas Westerbaan 10:22 Yeah. Well, you raise an important point because that's oftentimes, of course, a question that does come up in those discussions early on, really, is to say, well, I don't know this thing. Is it gonna be really big? Is this gonna be really intensive? Is it gonna require a lot of space? And is it gonna slow things down, especially if you think about response times on websites? People only have an attention span that maybe goes a few seconds. So if things slow down dramatically, then that would certainly not be a good thing. So maybe you can speak to your experience as far as, you know, how much or how little of an impact it had on your systems making that switch.
Johannes Lintzen 11:00 Yeah. So there's two things we worry about. I mean, performance is important. Right? Because customers, they look at benchmarks to see how far CDNs are, and they're not gonna pick one if we're very slow compared to the others. And experience matters. Right? It's not just if it's fast when we're sitting here with fiber Internet. It's also if you're on the plane Wi Fi board for the tour.
Bas Westerbaan 11:21 In those corner cases. Yeah. ks to see how far CDNs are, and they're not gonna pick one if we're very slow compared to the others. And experience matters. Right? It's not just if it's fast when we're sitting here with fiber Internet. It's also if you're on the plane Wi Fi board for the tour. In those corner cases. Yeah.
Johannes Lintzen 11:22 Yeah. So that's why we really try to get an idea of the performance here early on. I've already used a call back in 2019 with the first experiments with Google to get a sense of it. But that that's one thing, performance. I mean, that's important. And with Qualcomm Encryption, it seems that performance impact is manageable. It's not that bad. Right? In most cases, you won't notice. There's a tail where you notice, but it's good news mostly. The other aspects, which is the reason that things have took so long since twenty nine nineteen, is that it's not just about the form itself. Does it work in the first place? And even and TLS is crypto agile. Right? The protocol has been designed to be crypto agile. You can add new service to it, and it supports big things that is already sorted. That's the theory. But the practice is that not everyone implements the TLS standard as they should. And from our early investigations, we saw that some middle boxes, some software, if they see these bigger keys, even though they should just go negotiate old stuff, instead they break. They're like, oh, I don't know this. Right? So what they should do is they should respond with, I don't speak the queue. Let's just do classical crypto, which 95% of everything does and more than 95% does that. That. It's there was this small percentage, it's less than 5%, but it doesn't do this. It just breaks. Right? And we can't deploy something if 5% breaks or even 1% breaks. d do is they should respond with, I don't speak the queue. Let's just do classical crypto, which 95% of everything does and more than 95% does that. That. It's there was this small percentage, it's less than 5%, but it doesn't do this. It just breaks. Right? And we can't deploy something if 5% breaks or even 1% breaks.
Bas Westerbaan 12:44 So I think you're touching on two really important points, and one is implementation, right? Standards is one thing and it's great that we now do have standards, but implementation is what breaks. It's not necessarily right. It's the implementation that is creating the challenges. And then I think the other point that I want to pick up on is you used the term crypto agility, and I think that ties back into what you said earlier that really we're looking at several migrations, not just one. And the aspect of, yes, you have MLKEM for the encryption part and you have MLDSA and the other signature algorithms for the authentication part. So there's already two migrations on that journey. But then I think it's also crypto agility has often also been talked about the fact that we might not have another situation like we had with RSA and ECC, you know, fifty years of nice and very little changes just adding to the key length. The thing I sometimes struggle with is we're talking about crypto agility. And if you look into organizations today, the reality oftentimes is that they might be using Excel spreadsheets to manage their keys and certificates. So I guess from your point of view, from where you sit, how realistic would you say is achieving the state of crypto agility within today's companies, assuming that they follow your recommendations of getting executive buy in, doing their risk analysis, and coming up with a sort of a timeline within the constraints of their organization?
Johannes Lintzen 14:10 Yeah. So, I mean, crypto agility, it's maybe it's because I'm a technical person. What is actionable? Alright? And maybe to take a step back to get things bit concrete is, as you already mentioned, keys. If you have an organization, do you know which keys you have? And are you able to rotate all your keys within a day, week in case there was a big compromise? I think if you're honest, key management is today a very difficult problem already. And things that help there are automation, getting visibility, getting things in place. And that is difficult, but I think that's also part of it. If you want to do PQ, you also need to be able to do key management well in the focus. e? I think if you're honest, key management is today a very difficult problem already. And things that help there are automation, getting visibility, getting things in place. And that is difficult, but I think that's also part of it. If you want to do PQ, you also need to be able to do key management well in the focus.
Bas Westerbaan 14:52 Yeah. It goes hand in hand. Yes. That's for sure. We're talking a lot about encryption and authentication and security software solutions, applications. No. I suppose the biggest driver for anybody to really deploy those systems is, in the end, it comes down to to regulatory compliance. Right? So I think you, in the beginning, said that Cloudflare has started that journey already in 2017, '20 '18, sort of around the same time when this, I guess, started the process of finding those new algorithms. And then I suppose your release of the new standards in August of last year by NIST has really changed the conversations, at least that we're having. And I don't know, you might be seeing the same up until that point in time. We've been talking a lot about why haven't we been doing this? And like you said in the beginning, right, talking about quantum computers per se and the threat that they bring. But now that we have standards, it's almost like we have a starting line and we're talking much more about the, how can we tackle this problem? And maybe there are people out there that can already share from a background of experience how to tackle this. So I suppose from your perspective, what's your view on compliance and regulatory standards being available? Is is that more of an accelerator, or is it sometimes a bottleneck or a hindrance? I'd be curious to hear your take on that. e out there that can already share from a background of experience how to tackle this. So I suppose from your perspective, what's your view on compliance and regulatory standards being available? Is is that more of an accelerator, or is it sometimes a bottleneck or a hindrance? I'd be curious to hear your take on that.
Johannes Lintzen 16:16 That's a couple of questions. So, yeah, I think post quantum was on the radar quite a bit longer. There's been great work in academia with the bigger crypto workshop. But I think the moment that NIST started this competition, it was a clear sign that this will be required at some point. Right? I mean, they wouldn't do this at this point if down the line I mean, it's serious. What this pleasantly surprised me is how Dutch works is how thorough the US federal government has been in pursuing this migration. Their initial executive order then their post quantum bill, another executive order. They're really serious about this and about moving that. And I think that's a huge help in actually getting this on agenda, right, the interest there. And we see that there's other countries that are also actually differs quite a bit, but they're actually also some of them are quite in the front of thing in ahead of things. Germany is in quite well, France. I can't miss them all. But that really helps also the not just the compliance, but also the requirements that federal agencies have to prefer products that are PICU ready. If there's a choice, they have to go for the PICU ready thing, and that really helps the market move forward. Yes. So on the one hand, that's great. On the other hand, compliance can also be difficult in that compliance can especially with validation, FIPS validation and standards. Standards, you don't just have the standards of NIST itself. That's the downstream standard. They take time to write for people to get to put on paper. And tips validation, that's made for time when things move slowly. Right? It can take a few years before module is validated. And there's a tension there with the timelines that we have now. Although I also see that, at least within the federal government, they are changing things up a little bit so that you can be a bit more agile, so having validation afterwards. So there are there are things to do here, but there's certainly attention and which people are aware of. e with the timelines that we have now. Although I also see that, at least within the federal government, they are changing things up a little bit so that you can be a bit more agile, so having validation afterwards. So there are there are things to do here, but there's certainly attention and which people are aware of.
Bas Westerbaan 18:17 Absolutely. I agree. And I think, I mean, you bring up the federal executive order. I think there's been two that are really relevant for this migration discussion. Right? One was, I believe, in 2022, and then the one that follow-up was just recently released beginning of the year. And I think in that one, I mean, there's a lot of talk about AI and governance requirements for that, but then there's also a big section on post quantum migration. And and like you said, I think it spells out that MLKEM should be used whenever available. So I think you said it earlier, go talk to your vendors whenever you realize that you can directly influence the usage and the update of the key material algorithms that are being used. Go talk to your vendor and make sure you include that in your RFP, then ask them for support for the new standards. And I suppose a lot of this driven by the US federal government. Like you said, it's gonna have to trickle down and find itself in other standards. And I feel like just looking at some guidance that came out over the last couple of weeks, A lot of it was driven by payments, large payments organizations, FS ISEC and NCX9, even Europol, World Economic Forum. I mean, they've all sort of pushed out their guidance on this topic. So I think that's great to have that available. And that might also raise the awareness at the executive level where you say, okay, we hopefully gonna see more of those migration stories that hopefully repeat the success that Cloudflare has had. So I suppose given given everything that we spoke about, looking at just the immediate next year, the next twelve months, the folks, the organizations that are just waking up, as we said earlier, what's your recommendation on what should they focus on? What should their top priorities be in the short term, in the next twelve months to sort of get their organizations ready so as to ensure that they're not missing the boat?
Johannes Lintzen 20:05 You asked for the person who missed the boat.
Bas Westerbaan 20:07 But you made it nonetheless. Took a little longer. ould their top priorities be in the short term, in the next twelve months to sort of get their organizations ready so as to ensure that they're not missing the boat?
Johannes Lintzen 20:05 You asked for the person who missed the boat.
Bas Westerbaan 20:07 But you made it nonetheless. Took a little longer.
Johannes Lintzen 20:11 It took to see a detail. Yes. I mean, on that point, things can surprisingly take much longer than you expect. Right? And if we all wait until the last moment, there will be more work than we expect. And, also, are there resources available in the market at that point? Right? So I think it's good to take small chunks of this work. Right? Just get started with some things. I think that's important. Just get started. At the moment, we are in an interesting place that upstream software is already getting post quantum support. I think a lot of the migration will be quite famous. I mean, at some point, Windows will add PQ support. I mean, Chrome already has PQ support. You didn't have to do anything for that. Google did that for you. So a lot of this will be easy. A lot of migration will also be easy on your side. It's also important to start tackling the things that might be hard. Right? Get a sense of it. Try to experiment with some preliminary things. I think getting to know the work is half the work already, more than half the work already, probably. And maybe another approach is that it's perhaps tempting to see this as a problem which requires point solutions where you trickle it down in the organization and each team has to figure it out, and a little they have to solve the problem in their own small circle. Right? Whereas the reality of it is that cryptography is scary to touch. Right?
Bas Westerbaan 21:32 You expect it to function and be there, but be, like, somewhere hidden. Like the plumbing in your house. Right? It's a big mess and a big challenge if it breaks. But otherwise, you just expect it to be there and function.
Johannes Lintzen 21:43 Yeah. And then that cryptography is not completely and definitely broken. No one will touch it. I mean, why would you touch cryptography that's not broken. Right? I mean, it's a kind of a no one has been fired for not changing a cryptography. But but now we have to. And when we have to, I mean, the choice is not just change cryptography because also the way we use cryptography, we do it in a way because in a way that was possible. Right? Cryptography was small, so we have a lot of signatures everywhere. But the question is, do we need to do it at a higher level the way we have been doing it? Do we really want a drop in replacement, or does it work to perhaps do it in a different way? And I'm not talking just about the cryptography bits. Of course, it's true there as well. But it's also about just architecturally. If you're going to spend a lot of money for this change anyway, can't you change the way you do your IT infrastructure at the moment in a bigger way? Do we perhaps change vendor for the box? Do we perhaps I mean, there there's now that we have to do it, we can be a bit more. l. But it's also about just architecturally. If you're going to spend a lot of money for this change anyway, can't you change the way you do your IT infrastructure at the moment in a bigger way? Do we perhaps change vendor for the box? Do we perhaps I mean, there there's now that we have to do it, we can be a bit more.
Bas Westerbaan 22:45 Yeah. I think being able to follow a timeline, being able to draw from existing standards, regulations, recommendations, and pretty sure once this reaches critical mass, it it will also speed up the transition dramatically. Like you said, right, a lot of these things are gonna happen without any organization doing much. It's just gonna be there. But then there's other things that are gonna need a little bit more effort and proactive approaches to it. I think, Bas, this has been a fascinating conversation, so thank you very much for this. And I guess before we wrap up, is there any place you can point our listeners to where they can follow your work and learn a bit more about Cloudflare's post quantum initiative? Any blogs, anything that you would like to point out to our listeners?
Johannes Lintzen 23:31 Thank you so much for having me on the podcast. I feel we only scratched the surface. I mean, in a way, all cryptography will be post quantum cryptography in the future. Right? So, basically, we're just talking about how do we use cryptography and what does it mean for your organization? That's a really big discussion. Plus the whole change on top of it. Anyway, we have a lot written about both quantum and our experience on our blog, the CloudFlare blog. So blog.cloudflare.com, and then there's a post content tag that there's a lot of, I think, good stuff there.
Bas Westerbaan 24:01 I think it's really valuable information because you've done a big part of this journey already. So you're speaking from a background of experience, which is, of course, invaluable for someone who is just waking up as we said in the beginning. Right? So fantastic. Thank you for this, and I'm looking forward to speaking with you again. Thanks, Bas.
Johannes Lintzen 24:19 My pleasure.
Bas Westerbaan 24:21 The quantum era? It's not coming. It's here. But don't you worry. With the right tools, the right know how, and a can do attitude, we can tackle this thing head on. Want to learn more about the amazing folks at PQ Shield and how their top notch solutions can keep your precious data safe from those quantum critters. Mosey on over to pqshield.com. And, hey, while you're at it, why not search for Shielded, the last line of cyber defense on Apple Podcast, Spotify,
Johannes Lintzen 24:52 or
Bas Westerbaan 24:53 wherever you get your podcast fix. Don't forget to hit that subscribe button so you'll never miss an episode of quantum wrangling goodness. From all of us here at PQ shield, thanks for hitching a ride with us. Happy trails.