Latest / Technology Now / Can we build technology today to defend against the threats of tomorrow?

Transcript

Can we build technology today to defend against the threats of tomorrow? · Technology Now

• MICHAEL BIRD when you get a software update on your phone or your computer, are you the sort of person that will instantly update it or do you press the remind me tomorrow, until it gets to a point where your computer says no more and shuts itself down at a really annoying time? AUBREY LOVELL So I am definitely in the camp of instantly updating because I just, I'm excited to see what's new, what's updated. What about you? MICHAEL BIRD it will not surprise you that I also do the same because I also get excited about software updates. I think we can assume that software updates are pretty important. but I'm sure you've had it before, like I have, when you update something and you maybe get a new feature but it makes other things worse. AUBREY LOVELL we've all been there, right? Remember famously a few years ago, a very well-known company sent out an update and then we all had that blue screen of death. things happen, right? You try to make things better. There's updates and sometimes it goes completely horribly wrong. MICHAEL BIRD most of time you can just go back to the vendor and they'll sort out the problem. But what if the update is fake? What if it's malicious? So this week, Aubrey, we're going to be exploring how security is physically built into our devices to try and stop this happening. I'm Michael Bird. AUBREY LOVELL I'm Aubrey Lovell MICHAEL BIRD And welcome to Technology Now from HPE. MICHAEL BIRD So last week we explored the importance of post-quantum cryptography or PQC as the cool kids say. I think we're cool kids, Aubrey. AUBREY LOVELL think so. Let's hope. Maybe we're the only ones who think that. MICHAEL BIRD Yeah, so we looked at post quantum cryptography for governments looking to protect themselves and their citizens. But like almost every topic we discussed in the show, there is always more to learn. So let's look at this from another perspective. AUBREY LOVELL Alright, so in our P.Q.C episodes this year so far we’ve covered the new types of encryption needed to protect against the quantum threat and then last week we took a step back to look at the national importance of keeping everything secure against a quantum computer. So, what’s next? All right, so in our PQC episodes this year, this year far, okay, so far. So far away. All right, so in our PQC episodes this year, so far, we've covered the new types of encryption needed to kind of protect against that quantum threat. And then last week, we took a step back to look at the national importance of keeping everything secure against the quantum computer. So have to ask the question, what's next? MICHAEL BIRD because this week we are once again looking at security based around public and private keys, but this time we're considering how the private keys are kept secure because while they're private for now, if someone were to find a way to calculate them... then the technology they protect could be compromised. MICHAEL BIRD and it sort of relates to the thing, the question I asked you at the top of the show. So we're going to be catching up with Nigel Edwards, director of the security lab at HPE Labs later in the show to find out more about this. AUBREY LOVELL And I'm actually looking forward to this, before we get to Nigel, we're taking it way back. I mean, way back, all the way to ancient Mesopotamia because it's time for Technology Then. AUBREY LOVELL Well, we all know about secret recipes, right? Well, the concept isn't anything new because an ancient Mesopotamian tablet from 3,500 years ago and written in a language called cuneiform had a hidden recipe on it for a pottery glaze . Sounds interesting, right? I know, so fancy. But this tablet is the first known use of cryptography to hide secret information, and it used a very simple substitution cipher. So that's where one letter's kind of swapped out for another. So like imagine writing a book, but instead of the letter A, you would use the letter L and instead of B, you would use and so on, which is, yeah, that's a lot of work, but you know, they probably didn't have a lot to do back then, so. MICHAEL BIRD Aubrey, why, why, why, why is a pottery glaze recipe, why does it need to be kept secret? AUBREY LOVELL I don't know. Maybe it was a really good glaze. Maybe there was a competitor shops and like my glaze is better than your glaze. And you know, it's like your grandma's famous mashed potatoes, right? Like nobody's allowed to have them. And it's like guarded for generations, you know, in your kitchen. Exactly. But I mean, who knows, right? the first properly named cipher we know of is almost a millennium younger than what I'm going to call the Mesopotamian pottery cipher. And this is the Atbash cipher, an ancient Hebrew cipher dating back two and a half thousand years. We're really working through all the different centuries here. But like the pottery cipher, this was a simple substitution cipher, but it used a mirrored alphabet where the first letter was replaced with the last and the second with the penultimate and so on. So this style, so this style of encryption would be recreated yet again in what was known as the Caesar cipher. where letters are replaced by another letter in a fixed number of positions down the alphabet. I feel like this is turning into some like crazy movie right now, Michael, where we're like deciphering letters and it's an investigation. But the problem with all of these forms of encryption is that they are easy to break, right? So in fact, you can actually break them by hand using something called a frequency analysis or you simply look at how many times any letter appears. So in English, the most common letter is E, and once you work out which letter corresponds to E, you can very quickly decrypt something like the Caesar cipher MICHAEL BIRD yeah, I remember having little puzzle books when I was younger and they would have like little games like this where you have to try and decrypt AUBREY LOVELL Yeah, absolutely. as you know, right, like algorithms today really obviously are far more complex. So primarily because they need to be right. Modern computers could use AI to kind of tear that infamous enigma code to shreds in barely any time . And we've talked about that before on our podcast. So, you know, ancient substitution ciphers wouldn't really stand a chance. And, you know, the question is, how do we get around the issue of future technology being able to break today's methods of encryption? MICHAEL BIRD Well, Aubrey, that is an excellent question. And really is the question for today's episode because future proofing anything needing encryption is vital to modern security. So to that end, I spoke to Nigel Edwards. Nigel is the current director of the security lab at HPE Labs. we started the conversation by looking at how security is built into our IT estates. NIGEL EDWARDS I think what many people are unaware of is that if we take any of our devices, but if we say focus on a server, before a server can even boot the operating system and actually do useful work for the customer, there are a dozen or more firmware modules that need to be loaded into the various components that are in that server. firmware needs to be loaded into the processor, firmware needs to be loaded into the storage controllers, the network controllers and the management controller, in our case the ILO management controller, to bring that server into operation before it can even boot the operating system. MICHAEL BIRD So What sort of exactly is firmware? NIGEL EDWARDS So firmware is low-level software that provides management services and operates the components within, let's say, the server. So for example, a storage controller actually is running software, which we call firmware that enables it to communicate with the storage devices, get data from the storage devices, and pass that back over an internal bus, the PCI bus, in the server so that the processor can load it into system memory and then start to process on the data. MICHAEL BIRD So the thermos is an important component, particularly from a security perspective. NIGEL EDWARDS So this firmware is fundamental to the security of the platform. If an attacker could inject their own firmware into the system, potentially it gives them complete control over that platform- . They would be at a level below the operating system, below the level at which any system administrator interacts with the system, below any endpoint protection system. That would allow them potentially to spy on the customer and steal their data from other nefarious activities. MICHAEL BIRD Right, so what do we have in place to make sure you can't just load random firmware? NIGEL EDWARDS So today, firmware is signed by a private key corresponding to a public key which is fused or burnt into the silicon. So that means that the storage controller or ILO, the management controller for example, is able to verify the signature on the firmware before installs or loads that firmware. So we know it's coming from the legitimate provider of the firmware. For example, in the case of HPE ILO, the firm was coming from HPE and it's been signed by HPE. MICHAEL BIRD And so are those keys sort of quite similar to public and private keys when you're talking about, HTTPS or encrypting documents? NIGEL EDWARDS They're exactly the same. Algorithms are the same kind of keys. So typically we use the well-known RSA algorithm to sign firmware. So it's exactly the same algorithms that are used to secure the web, just a different application. MICHAEL BIRD And when you say burnt in, are we just sort of burnt in through software or is it physically a piece of hardware that's soldered onto the chip so to speak? NIGEL EDWARDS Yeah, it's a great question. So by burnt-in, it means the key is in a piece of immutable memory. It can only be written once, but it can be read many times. So during the manufacture of the chip, this memory is written. It's written with the key that's used, which is the public key that corresponds to the private key that's used to sign the legitimate firmware from the legitimate provider of that firmware. MICHAEL BIRD OK, that key's put on. mean, that sounds like that's perfect. No issues with that. But we have this thing called quantum cryptography. And that causes issues, right? NIGEL EDWARDS So if we take RSA, for example, the security of RSA is dependent on a mathematical problem which is known to be computationally infeasible to run on a classical computer. So in particular RSA, the RSA public key is a long number which consists of the product of two prime numbers. And if you've got the RSA public key on a classical computer, even if a supercomputer, recovering those two prime numbers would take you millions or even trillions of years. But if you could recover those two prime numbers, you can calculate the private key, so you have the signing key. Now unfortunately for us, there are algorithms which are known to be efficient to run on quantum computers which are very different models of computation from a classical computer. in a classical computer, a bit can have precisely two values, one or zero. A quantum bit can have potentially an infinite number of values and represent infinite number of states and that means it has a very different set of properties allowing us run different algorithms. So there's an algorithm developed by the American mathematician Peter Shor which efficiently will recover the prime factors for large numbers on a quantum computer. So instead of millions or trillions of years, potentially hours if a sufficiently powerful quantum computer is developed to recover the private key from the public RSA key. MICHAEL BIRD Okay, and so when we've talked about this in the past on the show, particularly with regards to software, the answer is, okay, you just rewrite the algorithm. So, you know, when we were talking about encrypting and decrypting files, you just use a quantum compatible method of encryption. But I suppose the problem here is if the method of encryption or the key is burnt into the device, then you can't change that key very easily. NIGEL EDWARDS That's right. once the key is burnt into the device, it's there for the life of the device. So that means the servers that we're shipping today, we cannot change the keys that are fused into the silicon for the operational life of that server. MICHAEL BIRD OK, so if a quantum computer suddenly appeared, then there are some serious vulnerabilities. And it's not just one vendor, it's every vendor. Everyone creating hardware right now has the same problem, right? NIGEL EDWARDS So all the components in a server or a switch or a storage array that load firmware need to be migrated to using additional quantum resistant algorithms in addition to the classical algorithms that they use. MICHAEL BIRD are they doing that at the moment or is it all just traditional? NIGEL EDWARDS So right now, HPE, as far as I know, is the first company in the industry to announce that they have our ILO management controller, which is using a new post-quantum algorithm to verify the firmware that it loads. But generally, No. The devices that are in our servers, our network controllers, for example, our storage controllers, these today are not using quantum resistant algorithms to verify the firmware that they're loading. So we're working, we need to work with our industry partners, our suppliers. It's not something any company can do on its own. call to arms within the industry. We need to migrate all components in all devices, switches, servers, and storage arrays to being able to using these new quantum resistant algorithms to verify any firmware that they load into these devices. MICHAEL BIRD So how do you know that a quantum resistant algorithm actually, how can you trust that that's going to work when a quantum computer suddenly turns up? NIGEL EDWARDS So the quantum resistant algorithms have been standardized by the American National Institute of Standards and Technology, but they're new. So we believe the maths is sound but there is a body of opinion within the community and also particularly from the European state security agencies that are advising that you can't, because these algorithms haven't been battlefield tested, we cannot discount the possibility that some cryptographer may discover a flaw in them or perhaps more likely that a flaw is discovered in the implementation. And so what that means is that because we need to be prudent when it comes to security, that at least for the short and medium term future, we need to run two sets of algorithms. A classical algorithm, such as RSA to sign the firmware, and in addition to one of the new algorithms, such as MLDSA, to sign the firmware. And then the device that's verifying the signatures has verify both signatures. Both signatures have to validate before that device will load or install the firmware. MICHAEL BIRD So it's a bit like double locking your front door. NIGEL EDWARDS I would say that's a great analogy. have to basically, are two checks that have to take place before firmware is authorized to load into the system. MICHAEL BIRD I mean, even if a quantum computer never turns up, surely that feels like quite sensible thing to do. You're just sort of bolstering the security. NIGEL EDWARDS Yeah, I agree with you. think it's a good thing to do. RSA has been around for many years. have been analyzing for many years. But the power of classical computers is improving all the time. And people are still developing new algorithms and new devices in the classical world, which may eventually lead to RSA being weaker than we currently believe it to be. MICHAEL BIRD so how soon do believe people need to transition to quantum resistant signing for firmware? And I mean, is it realistic that this is going to happen eventually? NIGEL EDWARDS Okay, so nobody really knows if and when a quantum computer... that can break RSA, and this is known as a cryptographically relevant quantum computer, will be built. So the Global Risk Institute has been surveying experts every year for last few years. So their most recent survey came out in December of last year. And two thirds of the experts that they surveyed said that by 2039, it was 50 % or higher probability that cryptographically relevant quantum computer would be available. We can't discount that there might be a major breakthrough and that it could occur sooner. So this is why state agencies around the world are now saying to everybody that you need to complete your migration to deploying post quantum algorithms by 2035. at the latest. But if you're running sensitive systems, then you really need to get it done by the end of 2030. So we need to make sure that all the components in our servers, in our switches, and our storage arrays are using quantum resistant algorithms as well as traditional algorithms like RSA. Because over time, it's likely towards the end of their life, even products that are built in the next few years will have to coexist and operate in the presence of a quantum computer. MICHAEL BIRD So do you think it could already be too late? you think there could be a cryptographically relevant quantum computer just in now? In existence and we just don't know about it NIGEL EDWARDS So our state-nation adversaries will certainly not announce to us that they are in possession of a cryptographically relevant quantum computer. So as far as we know, one doesn't exist. Certainly it's not thought that one exists in the Western world. MICHAEL BIRD I think I say that every single week, maybe it's a good sign, but I really enjoyed that conversation with Nigel because I think firmware is not like the most interesting of topics potentially, like it sort of has this sort of hidden implication, AUBREY LOVELL Definitely. What would you say is like your top takeaways from that conversation? I mean, there was just so much information, really good, you know, perspective as well on what could happen or what should happen. So I'm curious to hear your thoughts. MICHAEL BIRD I think the thing that I found most interesting, which is pretty obvious but actually when you think about it actually if an attacker can attack the firmware of a device, they would be at the level below the operating system an attacker could take hold of your switch or server or whatever it is that is running that piece of firmware. And you as the user could potentially have no idea that they have access to it because the operating system would sit on top of that and be completely unaware. So in terms of an attack vector, if you can get into the firmware, if you can inject malicious firmware, that is incredibly powerful. And I think the second thing is the fact that actually the advice is if you're in a running particularly sensitive environments, you need to migrate all of the devices that you're using to have quantum resistant algorithms in your firmware, So there are devices running now that could in theory be running in the presence of a cryptographically relevant quantum computer, which is worrying. Because I don't know about you, like I didn't realize that there were pieces of machines that were sort of physically unchangeable. I just thought everything could be updated with a software update, know, update the firm, update the software, but this is burnt in. AUBREY LOVELL we have so many devices, internet of things. Things are all interconnected and we swap them out It seems like it's just more susceptible nowadays to have something like that happen, go undetected and then really do that kind of viral damage in the background that you're not even aware of. It's almost like a game of chess, right, what your offensive, what your defensive, and always try to be two steps ahead of whatever possible outcomes there is. MICHAEL BIRD So I guess the big takeaway here is it's something that our organizations need to be thinking about. We may not necessarily be able to do anything right now because the hardware isn't there. But. We are getting there. AUBREY LOVELL think it's nice to feel like that assurance as well. know, like, it may never happen, but knowing that companies are taking it seriously and putting safeguards in place and thinking about, I think is super important. It's always in our best interest in technology to be proactive. So it sounds like a lot of those steps are being taken in the right ways. MICHAEL BIRD So Aubrey, as you know, I often like to ask our guests when they were younger what they wanted to be when they grew up, but, Aubrey, before we get to Nigel, can I ask you, what did you want to be when you grew up? AUBREY LOVELL You know this answer. Do you know what it is? I'm gonna put you on the spot. This is way, way back we talk about this. MICHAEL BIRD Can you give me a clue? AUBREY LOVELL what do I talk about all the time that I'm like obsessed with? And especially living in Florida, we have them a lot. MICHAEL BIRD You wanted to have something to do with storms something to do with hurricanes AUBREY LOVELL I did, yeah. I seriously considered it coming out of high school. What about you? MICHAEL BIRD it's all really transport related. I wanted to be a train driver when I was very young, as everyone does. But when I was older I seriously considered becoming a pilot MICHAEL BIRD Well, in case you're interested, producer Harry and executive producer Izzy wanted to a weatherman and a marine biologist, even though Izzy lived nowhere near the sea. Love that. But what I thought was nice about Nigel's answer is that his childhood dream job naturally turned into what he does now. Anyway, take a listen to this. NIGEL EDWARDS So as a 14 year old, the physics that I was studying wasn't very interesting. And then when I got to about 18 or 19. That was just fascinating. So the kind of whole physics and the mathematics. The principles of flight. And all those principles that were becoming, you know, it suddenly got really interesting to me. So I decided to, I then had great difficulty in choosing between physics and engineering. So I chose engineering. But yeah, the 14 year old Nigel wanted to become a pilot. The 18 and 19 year old Nigel found mathematics and physics and engineering much more interesting and wants to study at university. AUBREY LOVELL Okay that brings us to the end of Technology Now for this week. Thank you to our guest, Nigel Edwards, And of course, to our listeners. Thank you so much for joining us. MICHAEL BIRD If you’ve enjoyed this episode, please do let us know – rate and review us wherever you listen to episodes and if you want to get in contact with us, send us an email to technology now AT hpe.com, subject line: Mesopotamian Tablets, and don’t forget to subscribe so you can listen first every week. Technology Now is hosted by Aubrey Lovell in sunny St Petersberg, Florida, and myself, Michael Bird just outside of London, where the weather is also sunny and really, really hot and humid. Not very nice. This episode was produced by Harry Lampert and Izzie Clarke with production support from Alysha Kempson-Taylor, Beckie Bird, Allison Gaito, Alissa Mitry and Renee Edwards. AUBREY LOVELL Our social editorial team is Rebecca Wissinger, Judy-Anne Goldman and Jacqueline Green and our social media designers are Alejandra Garcia, and Ambar Maldonado. MICHAEL BIRD Technology Now is a Fresh Air Production for Hewlett Packard Enterprise. (and) we’ll see you next week. Cheers!