The Future of Cloud Native with Arm’s Cheryl Hung at KubeCon Paris 2024
At KubeCon Paris 2024, Mitch Ashley and Cheryl Hung discuss how Arm is powering the future of cloud native AI, and the cloud native community.
Transcript
This is Textron tv. Hey everybody. Mitch Ashley here at Kku Con in Paris 2024.
Taking over a little bit for Alan. He's been doing some interviews and I'm jumping in and meeting with some fantastic people, one of which is Cheryl Hung, who's with ARM Work. Welcome, Cheryl.
Thank you so much. It's fantastic to be Here. Good to have you.
Tell us about what you do at arm. So, Adam Arm, I'm the senior director of Ecosystem within the infrastructure line of business. Mm.
So that means arm in data centers, cloud, telco, networking, That, and that's really growing. I mean, I, I see more and more ARM in the data center, arm in the cloud. It's, it's the fastest growing part of arm.
ARM is very dominant in mobile phones and mobile devices already. But Cloud and of course ai, the whole topic of this, this conference, it feels, um, are huge drivers for arm. Very big.
So I'm really happy to be here. I'm curious, uh, so you, you have some connections back though into CNCF before you were with arm, right? Indeed, indeed.
I actually used to work for C-N-C-F-A few years ago. Um, I used to lead the end user community as the VP of ecosystem in CNCF. So I care a lot about, you know, not just good technology, but the community, making sure that it's very easily accessible and very usable.
Um, and supporting people from whatever big or small organizations or individuals to come and be part of the Community's nice to have that continuity when you move to harm and bringing all those relationships and that knowledge, that experience with you. Uh, it is definitely, like, one of the things I love about this community is we're all friends here. You know, it doesn't matter where you work, doesn't matter what you are doing today, but the, the relationships that people form here have lasted, you know, beyond what their company's doing.
One company and, you know, they really maintain that. So every Great coupon, at least one person says that to me. Yeah.
You know, so I think it is really true. It is a community. I mean, the first coupon that I came to was Berlin in 2017.
I'm gonna say it was 1500 people. I was here on a diversity scholarship, so I knew nobody, you know, much smaller conference, obviously a lot of energy and buzz, but at the time it was still like, oh, Kubernetes and maybe a handful of other things. Yeah.
And now you look at it at Kubernetes conference to really being Yeah, exactly. Cloud native Distributed ommunity. Yeah.
Now Kubernetes is so established and, you know, solid. And that at the time it was still like, oh, is Kubernetes even gonna be a real thing? So I'm very, very ratified, you know, to see how this community's grown.
What, so three things, three things I'd love to talk with you about. I'd love to hear more about how ARM is doing in the cloud. 'cause I, more and more I hear those kind of conversations happen on the infrastructure side.
Um, open source project adoption of using arm, and then also of course ai. You know, we, you can't go five minutes into a conversation without talking about ai, but great. Legitimately, I mean, with ARM and AI projects.
So choose the order. You could talk about whatever you want. Let's, um, let's go from, let's go from the top.
Okay. So, um, last coupon Oracle announced that it was offering millions of dollars in cloud cloud credits for ARM servers. So offering these two CNCF projects.
Um, and the goal here is really to enable CNCA projects to experiment with ARM servers, to test, to deploy. And the goal is really for all of the projects, eventually to be multi architecture. So independent of what your end users are using, you know, today you should be neutral towards what architecture and you should try and support a broader range of end users as you can.
Mm-Hmm. So this announcement that we made with Oracle was really a way for us to show that this is something that we want our community to support and to take advantage of. So part of my goal of being here today is to talk to people and tell them like, oh, what are you doing with these cloud credits?
Can we help you? Can we do anything for you to make it easier for you to access those? Um, so yeah, that's, that's one of my main, so, so How's it gone since the announcement?
How, how's that adoption happening? What do you see happening with things? Its picking up.
It's picking up, yeah. Yeah. The, the challenge of course is um, is obviously a smaller player.
Mm-Hmm. So far in the cloud native space, it's This thing called momentum too, right. You Know, momentum.
Exactly, yeah. Gathering, you're gathering, uh, more and momentum as should Do this. Exactly.
And that actually takes me to the second topic you mentioned, the open source adoption. We are very, very proud to say that more than 80% of the graduated projects support arm, that is huge. That is not where we were three years ago, and we would love to get that to a hundred percent.
So the maintainers, the TOC, you know, we are really talking to members of the community on how we can get that number to a hundred percent. Um, part of that. So, you know, supporting through things like cloud credits, part of it's supporting through supporting the maintainers and the community directly.
Um, but it's a significant achievement. Bear in mind, ARM is not a software company, right? Right.
Yeah. A few years ago, ARM is like purely in chips and computer hardware. So for ARM to actually make the stance and say, you know what?
The software matters and when it comes to software, open source software really matters. That is a big change for arm, a big cultural mind shift. Um, so I'm very, very proud of the company for that.
Let me ask you then, so is it, is it just, is it, I'm gonna be simple, simple answer. You tell me if it's the simple, I'm, I'm guessing it's not. Is it just getting projects to distribute their binaries in ARM and Intel and whatever, you know, chip formats?
Or are there more things about how you incorporate design into your software to take advantage of different use cases for arm? So, both. Both, I would say, um, on one hand, moving recompiling to ARM is actually very simple.
Um, arms put a lot of effort into making the compilers, you know, you basically flip a switch and in theory it should all work. Of course, that, you know, last little bit is where all the effort is to make it just work. There's always so anything like that.
Sure. Um, so there is a lot of work still left in terms of testing and making sure that software works together in the way you would expect it to. Mm-Hmm.
Um, so, you know, I'm not saying that it's an easy task. There's definitely still work to do there. And the second part is on particular workloads.
So of course the, the hot topic of the day is ai, and AI is driving a lot of the demand for arm, um, an arm in the cloud. In fact, I gave a keynote a different conference, um, last week where I said, if you're not on the AI hype train, you need to get on it now. You know, I was a little bit of a cynic myself, but, you know, seeing the momentum here, seeing how, seeing how AI is really being used to solve these problems that would've been too difficult or too expensive before.
Um, it just goes to show, you know, that's where the, that's where the hype trainers and obviously, um, you know, in collaboration with some other players like Nvidia is a big part of that. So we are doing our best to keep pushing, keep supporting that, um, community, and to democratize the access to GPUs and make it easier for people to use them and run them in production. Great.
Are there, are there certain things you need to work with different LLM providers to help them work in an armed environment? Is that important at all, or that's kind of a little, a little higher up in the, in the stack. That's just not as big issue.
We, a surprising amount of it just works. You know, a lot of it is really, really, um, you know, just works. And that's really a tribute to all of the closed and open source, you know, engineers who've been working on this, um, over the last few years.
There are obviously some things, not so much in, you know, the data, but in the training and the influence that maps need, uh, need some work and need some polish. But honestly, honestly, the, the biggest challenge right now is all of the layers above it. So things like Kubernetes, we saw from the keynotes today that actually there's still a lot of challenges in how you manage those GPUs and share those resources effectively and fairly and, you know, cost effectively from a business side, um, and keep your users happy because obviously they want, you know, the fastest access possible.
And so there's an absolute ton of those. And I say the other big bit that we are missing right now is actually in the ecosystem. You know, we don't have necessarily, the talents policies are changing really fast.
Mm-Hmm. Governments are figuring out what to do about this whole AI thing. Um, and the hardware is still difficult to access, very Expensive on all parts of it.
It, it seems like, I mean, AI, ML'S been around for a while, but especially Journey AI is all they do. Most of us, You know, it's, it's, it's been around, but it's been around for the very big tech companies. Right.
You have the deep pockets. If we're gonna get that out to the widest range of people and companies possible, then we need all these things. We need the money, we need the talent, we need policies to support it, and we need the hardware to be accessible as well.
So we definitely have a long way to go, but it's really great to be at the beginning of the journey and, you know, see where we land in five years. Yeah. You do kind of have to be there at the start, you know, to really be part of the whole ecosystem.
How, how about, I'm curious, you know, there's a lot of discussion about, uh, demos, domain specific LLMs, um, geographically based LLMs that are at the edge. Now we're talking about just gen AI at the moment. It seems like those are, you know, really particularly great use cases for ARM as well as, you know, more efficient user resources, sustainability Yeah.
As a big issue around ai and it's, it's, it's kind of elevating sustainability even more as a topic than it already is. Yeah, 100%. Um, ARM has always had this focus on, um, being energy efficient right from the beginning.
That was, you know, the goal was these very low energy chips and that served it very well in mobile devices where, you know, you want to have your battery live for your phone last all day, but turns out that's also wonderful if you have massive data centers and you are using an enormous amount of, of energy Mm-Hmm. Then using that energy effectively is super important. So many businesses, you know, almost every business I'm talking to has sustainability as one of its long-term goals.
And one of the best ways to do that is to make sure that you're using your energy and, you know, it's also good for the company, right? Because you energy costs money, you're spending a lot on your cloud resources, so you wanna get the most out of them. There's good, good for the company, good for the planet, and good for people.
Excellent. Yeah. Well, uh, so wrap up with kind of more of a, a you question as a veteran, coming back to Kup Con, having gone multiple years, multiple events, is there any particular meetup or group or event or part of kon, like is the top of the list of, I always show up at the Wednesday, whatever, you know, is there something like that for you that you really enjoy participating in the most?
Um, I would say on a personal level, um, I run the cloud Native London Meetup. I've run it since 2017, so gosh, seven years now. I've been running this meetup.
We have over 8,000 members, and I always love to meet the meetup organizers from different countries. Mm-Hmm. So, you know, the one in Paris, the one in the Nordics, you know, the one all over.
And I, I love meeting up with these folks because they're always really passionate about getting people together, you know, learning new tech, sharing what they know. So for me, I think getting in touch with the Amba other ambassadors is key. And then, yeah, just hanging out with people, you know.
That's great. I think it's always good to meet people you don't know as well. It's, you meet people you don't know and, and you know, it's, we're Covid iss somewhat in the rear view mirror for, for us, but I'm still meeting people.
I've only met online and like, yeah, just, just someone walked up today and like, Hey, I finally get to meet you in person. We're still making that personal connection. Just two minutes ago I saw someone I knew and I was like, oh, after this is done, I might have to catch up.
I gotta go Try and catch them. Exactly. Well, it's been fantastic talking with you, Cheryl, and, uh, we, you the best and I'm excited what ar what Arm is doing in the industry and the adoption level.
You know, I have one more question if I get ask You. Yeah, go for it. I Have this theory, I don't know how big of an impact it has, you know, being developer in my background and, and long, long term Apple users, when Apple moved over to an arm style chip, at least that instruction set, it seemed like that kind of brought a whole new developer community into the arm world needing ARM software, all the things you're talking about Of course.
Which helps when you move it into the cloud and into the data center. So it seems like moves like that where developers are working and the software they're working on is seems, seems like one of the things that will help accelerate that. Is that true?
It is very true. It's, it's actually a little bit embarrassing how often I hear, you know, oh, I moved my company to Arm because I wanted to use my Mac with my fancy new M two. Yeah.
That's what gets them there, right? I'm like, sure, yes. You know, that's, that's great.
Um, I actually used to work for Apple before Arm, so I've also seen that from both sides. Um, but you know, it just goes to show that, you know, arms growth is everywhere and it's wonderful and it, it also really goes to show how developers are leading the charge. Mm-Hmm.
You know, a lot of things, you know, like, yeah, okay. Sustainability might not be something the engineers care about day to day, but they care about their battery life. So we're all heading in the same direction.
Excellent. Well, Cheryl, it's been a lot of fun. Cheryl Hung, who's with Arm, you wanna know something about what's happening in the cloud or CNCF or Apple or Arm.
Not to, not to overload too many expectations on you. It's been great talking with you. It's been Wonder appreciate you, you coming by.
Been wonderful talking to you. Thank you so much, Mitch. You Too.
Have a great show. Thank you everybody. We will be right back with our next fantastic guest, just like Cheryl.
And I hope you stay tuned. We're just getting started here at K Con in Paris. See you in a minute.