KubeCon + CloudNativeCon Europe 2024 | Sudha Raghavan | Oracle

Clips
News
More from KubeCon + CloudNativeCon Europe 2024

Sudha Raghavan

SVP Oracle Cloud Infrastructure (OCI) Developer Platform and CNCF Board

Oracle

Oracle’s contributions to the CNCF and a growing open-source landscape

While it’s mostly known as a player in the integrated cloud applications and platforms landscape, Oracle Corp. also has a strong open-source community presence.Having participated in over 500 projects over the years, the company has been steadfast in its commitment to community engagement.“The biggest part of all of this is we’ve been doing it for decades,” said Sudha Raghavan (pictured), senior vice president of the OCI developer platform at Oracle. “It’s just that we haven’t been very public about it. We’ve always done Oracle, Linux and Java, but now we are trying more and more [Cloud Native Computing Foundation] projects. We are becoming one of the planning sponsors for CNCF

play_circle_outline Oracle's Transformation: A Commitment to Open Source and CNCF Platinum Sponsorship

play_circle_outline Oracle's focus on growing the community of women in tech

play_circle_outline Advice for women entering the tech space, especially in Kubernetes

Info

Sudha Raghavan | Oracle

Sudha Raghavan

SVP Oracle Cloud Infrastructure (OCI) Developer Platform and CNCF Board Oracle

search

(bright music)

>> Good Morning, cloud community, and welcome back to fantastic Paris, France. We're here at KubeCon, CloudNativeCon, CNCFs flagship European event and the largest KubeCon ever with over 12,000 attendees. My name's Savannah Peterson joined by masterful analyst, Rob Strechay. Rob, we've had the best time here in Paris.

>> I think this has been one of the best KubeCons I've been at. I mean, again, it's hard to do it wrong in Paris, but I think that just the energy, 51% new people here, new energy coming into the building, it's been awesome.

>> It really has been awesome and speaking of awesome, our next guest is super awesome, Sudha. Thank you so much for being here with us. It's been a busy week for you, yeah?

>> Yes, extremely busy, but exciting.

>> Yeah, and exciting, I just got to point out that fabulous sequin bling on your Oracle shirt, loving that.

>> Thank you.

>> You rocked that on the main stage for your keynote today, correct?

>> Yes.

>> Yes, congratulations on that.

>> Thank you, again.

>> How did it go? Can you give us a couple highlights?

>> So the keynote was great. The biggest thing we wanted to get out there is that Oracle is a huge sponsor of open source. Not only do we use open source, we also contribute back.

>> Over 500 projects, right?

>> Yes, 500 open source projects. and the biggest part of all of this is we've been doing it for decades. It's just that we haven't been very public about it and now we're coming out party, so we're telling the world what we are contributing to. We've always done Oracle, Linux, Java, but now we are trying more and more CNCF projects. We are becoming one of the platinum sponsors for CNCF. I am part of the CNCF board and it's an exciting phase to be part of Oracle and see this journey through from sort of the Oracle DB that we are always known for to the Oracle Cloud enterprise that we are now known for.

>> So why is the coming out party happening right now? I mean, I did not realize until I was prepping for you that Oracle had contributed to over 500 projects. It's extremely impressive, what made this moment the time you all wanted to talk about all of these?

>> It's the community, right? You guys spoke about it, right? There are 12,000 people out here and this is the developer community. Who do we talk to about our contributions to open source? Not the CIOs and the CTOs, but the developer community and this is the perfect conference for it.

>> Well, I think also part of it was that when people think Oracle, they don't necessarily, they think database a lot of times. So what's your pitch to developers to come to OCI? I liked how you talked about how you were basically building all your services on Kubernetes going forward and I think that to me shows your investment in the community and why you're here, but what's your pitch to developers to come and develop on OCI?

>> So OCI from the very get go, we are the generation two of cloud. So we have the late movers advantage, and what does that mean? That means we go to the customers where they wanted us to be. Customers don't want rental lock-in. They want to be cloud ubiquitous. I code to an interface. Run me in any cloud that allows that interface to run seamlessly. That's exactly how we built our products. Everything is open standards compliant. We are not only building our services to be open standards compliant. Being part of the board now, I get the voice to be able to bring standards to other things that are not yet standardized. For example, our GPUs that are in the highest demand.

>> Yeah, I thought that was an interesting one 'cause we're talking about AI, but go ahead. You looked like you had something there, but.

>> I'm just smirking because I mean, the way that you casually say that, right? We were at Supercomputing in Denver earlier or at the end of last year and one of my favorite shirts from there was "Got GPUs?" But that's exactly, you're bringing up a really good point about standardization and how that allows everyone to accelerate their development and build faster, create more. Do you think we're going to see a lot more standardization with hardware and as we enter into AI?

>> Absolutely, it is all about evolution, right? 10 years ago, there wasn't Kubernetes. Today, that isn't a world without Kubernetes, right? This is what it's going to look like. AI has to become democratized. Can't have one or two vendors supplying it to their custom specifications that all of the data scientist community, all of the developer platform community has to adopt to and individually tune. That's never going to scale. We need something a little bit more standardized so you can again, write once, run everywhere, right? How do we make that happen for the AI software? It needs to become commoditized.

>> I thought also it was really great and we have a power law of GenAI, and what we see is that you on the top part where it kind of slopes down, you have all of the big models that are being trained in the cloud and the foundation models. And then in the middle, what's tugging out the middle is open source like the open source models. We heard a lot from Llama this week and you have Meta with Llama Two, with LLama Three coming.

>> Lots of llamas.

>> Lots of llamas, llamas everywhere, but what was also interesting, and I thought you hit on it in your keynote and it's been hit on a couple times with Ampere and ARM and what you're doing there about actual AI workloads running on CPUs because for us, the long tail is what we would call SLMs, or segmented, or small language models that are being deployed at the edge or in other instances. How do you see that developing?

>> So you spoke about with the highest end where you actually need the power of the GPU. So one of the things that I did not talk about in the keynotes is we at Oracle are trying to build the largest GPU cluster in the world. We are going to build a cluster for 16,000 nodes all running again on Kubernetes, right? So and we are very close. We are almost halfway there. We'll be there in a few weeks. That's the highest end, that's exciting, right? It's one of a kind, it's the first of the kind, and we have workloads that need that cluster. But there are also workloads like inferencing the smaller models where innovation can come really fast as long as we can supply the infrastructure to help with that innovation. That's why we tried CPUs and it works, right? We have proven data that it actually works equally or sometimes even better depending on the workload to run the same inferencing model on CPUs. And what you were saying is so true because inferencing is the long tail. You train and you fine tune and you inference. Yes, you do feedback, but inference runs the longest. So again, if we pivot towards running the longest running piece of this workflow in something that is a little more ubiquitous, a little more available, than best of both worlds.

>> Yeah, I think that to me comes back to and again, you're heading up all of the developer aspects within OCI, I think what's really interesting is and you kind of hit on it a little bit, and there's some interesting subtitles going on here around infrastructure as code and how that's going to developing, and where things like OpenTofu and we'll see if the news of Hashi potentially selling or being bought, how that changes things. Where do you see some of the standardizations going with infrastructure as code beyond just Kubernetes, I mean, as a foundational model?

>> So that's an interesting area of development. Really OpenTofu is a great example, right? As soon as HashiCorp decided whatever it was going to do with Terraform, OpenTofu came into life. That happened almost seamlessly. It's because both like CNCF. There's always somebody, someone ready to make what is now private open sourced and that's the great thing about the community. You spoke about what do you think about infrastructure as a code standardization? I go back to open standards, right? Now only if your infrastructure is written to a standard specification and in the old days, they called it terraform providers. I'm calling it old days as though it was decades ago. It was just a year ago, but.

>> In tech land, a year is a decade so.

>> Dog years they were saying in Kubernetes, right? But that standardization allowed us to write code for the infrastructure. If I had gone to someone 10 years ago and said, "You can deploy on an Intel machine the same way you deploy on an AMD, the same way you deploy on a Mac," they'd go, "What, really? You can actually do it." And now people don't even think about it. A lot of my customers are what we call shape agnostic. They don't know what hardware is running. They just know that containers are running to the scale that they want to and scale back down to zero. They don't care what infrastructure you run on. As long as it's cost efficient, really nobody cares and that is what standardization brings to the world. And not just that making infrastructure so ubiquitous means the developers, most of who are trying to bring value to their customers, don't have to worry about infrastructure. It's there, it's always there. It is standard, it's scales on demand. What more can the developers ask for? So standardization on infrastructure as code is definitely coming and we are at the forefront of doing that. Not just for the traditional cloud workloads. We're trying to do it for the AI workloads too. Acute flow is a great example where started democratizing what data scientists do which wasn't a thing again, a few years ago. But now everybody uses it.

>> Right, and bringing it to the platform engineering aspect of it as well, so yeah.

>> Which has definitely been a theme. So you're at Oracle, you're on the CNCF board, and you are also a part of the Oracle Women's Leadership Seattle Chapter, founding member, correct, which is super awesome. Do you sleep is my initial question here because you wear many hats?

>> I try to.

>> Nap occasionally, periodically when the moment strikes. So how are you and Oracle helping grow our community of women in this space 'cause we're still pretty outnumbered?

>> Yes, so what's been great about coming to conferences like this? First thing that when I came to this conference, Wednesday morning, 8:00 AM, there was a woman's reception that happened. I'm sorry, I forgot the sponsors, but I met a woman who was a PhD in physics, physics, and she was here to learn Kubernetes, fascinating.

>> I love that.

>> I met people from marketing, from sales, from development, so women are everywhere. It's just we don't seem to grow the women from the bottom of the stack, the funnel coming in. That was a problem. We've gotten better at it. It's not great, but we've gotten better at it. Now we need to focus on the middle tier to get them a little bit higher and primarily I try to do that with a lot of mentoring. I am a huge fan of mentoring as a concept. I am not so great at networking, but mentoring, I love. I have benefited from being a mentee to so many different men, women, no difference. But that has propelled me in directions that I would've not thought of without them sort of being there as friends and guides to take me to that next step. So I like to give that back to the community and I do a lot of mentoring myself. Here I met some wonderful women. Again, the Priyanka who did the keynote.

>> Yes, yeah, we had her on the show on Wednesday, yeah.

>> Yeah, and you can see CNCF board, not just the board, the CNCF Foundation is only about 50 people but they are over 30% women. And that's great, they built that on purpose. I was talking some other gentleman yesterday and he was saying we don't need to do this as women and men. What if we take the gender out of this equation? Just evaluate us for who we are, what experience we bring in, and let's go from that basic sort of neutral boundary. and then we will see women shine as much as men do. And maybe more, I wear a lot of hats.

>> Yeah, yeah, yeah, well, I think what you're bringing up is really important and it's not just about that gateway. It's about elevating women into leadership and making sure that they feel empowered to take that step. What would be your advice to a woman who might be watching this right now who knows nothing about Kubernetes like the woman you met who has a PhD in physics? What would your advice be to that person about entering the space?

>> So take the trainings, right? Everything today, in this world where data is in our fingertips, we get a lot of data. But it's my true belief that when you take that first action from the data that you get, no matter where you hear it from, that's when it really sticks and what's the worst that can happen? You can say this is not for me, but now it's not that nagging little thing in the back of your mind saying, "I should have tried it, I should have tried it, I should have tried it." So take the action, go to CNCF, look at the training materials. There's so many free trainings out there. Try to set up your own Kubernetes cluster. If nothing else, really I don't know how many people watched the keynote this morning. Some of the biggest contributors now to Kubernetes, they started their first contribution was changing a line of comment. Somebody said their first contribution was changing a legal contract. You don't expect them from those kind of comments, first comments into Kubernetes from people who are very well known in the community, but that's where they started. So do something about that stack.

>> Start small.

>> Absolutely.

>> I think that's really great. All right, last question for you 'cause this has been a fascinating interview. What do you hope that you can say next time you're sitting here on the show with us that you can't say today? So that could be in Salt Lake or London. What do you hope happens in the next six to 12 months?

>> The acceleration of AI-driven development for the betterment of humankind.

>> Rob: Love that. I just got goosebumps, yep.

>> I am a huge proponent of climate change. I really truly believe that if we don't take action today, tomorrow is going to be really hard. You live on the west coast of United States, so do I. We've seen the number of wildfires every year grow up and up and up, and it's real. It's affecting us, everyday lives, and if you don't do something today, it's never going to get done. So I hope six months from now I come back and I see AI driving that change.

>> Well, I hope we can have that exact conversation 'cause I'm here for it in Salt Lake City when we're at KubeCon there. That would be fantastic.

>> I'll be there too.

>> Oh, see, so we're going to do it. Hopefully we're saving the world at that point. Sudha, thank you so much for being on the show. You're absolutely fantastic, Rob, a pleasure as always. And thank all of you for tuning in for our live coverage here at KubeCon, CloudNativeCon in Fantastic Paris. My name's Savannah Peterson, you're watching theCUBE, the leading source for enterprise tech news. (bright music)

Sudha Raghavan | Oracle

search

(bright music)

>> Good Morning, cloud community, and welcome back to fantastic Paris, France. We're here at KubeCon, CloudNativeCon, CNCFs flagship European event and the largest KubeCon ever with over 12,000 attendees. My name's Savannah Peterson joined by masterful analyst, Rob Strechay. Rob, we've had the best time here in Paris.

>> I think this has been one of the best KubeCons I've been at. I mean, again, it's hard to do it wrong in Paris, but I think that just the energy, 51% new people here, new energy coming into the building, it's been awesome.

>> It really has been awesome and speaking of awesome, our next guest is super awesome, Sudha. Thank you so much for being here with us. It's been a busy week for you, yeah?

>> Yes, extremely busy, but exciting.

>> Yeah, and exciting, I just got to point out that fabulous sequin bling on your Oracle shirt, loving that.

>> Thank you.

>> You rocked that on the main stage for your keynote today, correct?

>> Yes.

>> Yes, congratulations on that.

>> Thank you, again.

>> How did it go? Can you give us a couple highlights?

>> So the keynote was great. The biggest thing we wanted to get out there is that Oracle is a huge sponsor of open source. Not only do we use open source, we also contribute back.

>> Over 500 projects, right?

>> Yes, 500 open source projects. and the biggest part of all of this is we've been doing it for decades. It's just that we haven't been very public about it and now we're coming out party, so we're telling the world what we are contributing to. We've always done Oracle, Linux, Java, but now we are trying more and more CNCF projects. We are becoming one of the platinum sponsors for CNCF. I am part of the CNCF board and it's an exciting phase to be part of Oracle and see this journey through from sort of the Oracle DB that we are always known for to the Oracle Cloud enterprise that we are now known for.

>> So why is the coming out party happening right now? I mean, I did not realize until I was prepping for you that Oracle had contributed to over 500 projects. It's extremely impressive, what made this moment the time you all wanted to talk about all of these?

>> It's the community, right? You guys spoke about it, right? There are 12,000 people out here and this is the developer community. Who do we talk to about our contributions to open source? Not the CIOs and the CTOs, but the developer community and this is the perfect conference for it.

>> Well, I think also part of it was that when people think Oracle, they don't necessarily, they think database a lot of times. So what's your pitch to developers to come to OCI? I liked how you talked about how you were basically building all your services on Kubernetes going forward and I think that to me shows your investment in the community and why you're here, but what's your pitch to developers to come and develop on OCI?

>> So OCI from the very get go, we are the generation two of cloud. So we have the late movers advantage, and what does that mean? That means we go to the customers where they wanted us to be. Customers don't want rental lock-in. They want to be cloud ubiquitous. I code to an interface. Run me in any cloud that allows that interface to run seamlessly. That's exactly how we built our products. Everything is open standards compliant. We are not only building our services to be open standards compliant. Being part of the board now, I get the voice to be able to bring standards to other things that are not yet standardized. For example, our GPUs that are in the highest demand.

>> Yeah, I thought that was an interesting one 'cause we're talking about AI, but go ahead. You looked like you had something there, but.

>> I'm just smirking because I mean, the way that you casually say that, right? We were at Supercomputing in Denver earlier or at the end of last year and one of my favorite shirts from there was "Got GPUs?" But that's exactly, you're bringing up a really good point about standardization and how that allows everyone to accelerate their development and build faster, create more. Do you think we're going to see a lot more standardization with hardware and as we enter into AI?

>> Absolutely, it is all about evolution, right? 10 years ago, there wasn't Kubernetes. Today, that isn't a world without Kubernetes, right? This is what it's going to look like. AI has to become democratized. Can't have one or two vendors supplying it to their custom specifications that all of the data scientist community, all of the developer platform community has to adopt to and individually tune. That's never going to scale. We need something a little bit more standardized so you can again, write once, run everywhere, right? How do we make that happen for the AI software? It needs to become commoditized.

>> I thought also it was really great and we have a power law of GenAI, and what we see is that you on the top part where it kind of slopes down, you have all of the big models that are being trained in the cloud and the foundation models. And then in the middle, what's tugging out the middle is open source like the open source models. We heard a lot from Llama this week and you have Meta with Llama Two, with LLama Three coming.

>> Lots of llamas.

>> Lots of llamas, llamas everywhere, but what was also interesting, and I thought you hit on it in your keynote and it's been hit on a couple times with Ampere and ARM and what you're doing there about actual AI workloads running on CPUs because for us, the long tail is what we would call SLMs, or segmented, or small language models that are being deployed at the edge or in other instances. How do you see that developing?

>> So you spoke about with the highest end where you actually need the power of the GPU. So one of the things that I did not talk about in the keynotes is we at Oracle are trying to build the largest GPU cluster in the world. We are going to build a cluster for 16,000 nodes all running again on Kubernetes, right? So and we are very close. We are almost halfway there. We'll be there in a few weeks. That's the highest end, that's exciting, right? It's one of a kind, it's the first of the kind, and we have workloads that need that cluster. But there are also workloads like inferencing the smaller models where innovation can come really fast as long as we can supply the infrastructure to help with that innovation. That's why we tried CPUs and it works, right? We have proven data that it actually works equally or sometimes even better depending on the workload to run the same inferencing model on CPUs. And what you were saying is so true because inferencing is the long tail. You train and you fine tune and you inference. Yes, you do feedback, but inference runs the longest. So again, if we pivot towards running the longest running piece of this workflow in something that is a little more ubiquitous, a little more available, than best of both worlds.

>> Yeah, I think that to me comes back to and again, you're heading up all of the developer aspects within OCI, I think what's really interesting is and you kind of hit on it a little bit, and there's some interesting subtitles going on here around infrastructure as code and how that's going to developing, and where things like OpenTofu and we'll see if the news of Hashi potentially selling or being bought, how that changes things. Where do you see some of the standardizations going with infrastructure as code beyond just Kubernetes, I mean, as a foundational model?

>> So that's an interesting area of development. Really OpenTofu is a great example, right? As soon as HashiCorp decided whatever it was going to do with Terraform, OpenTofu came into life. That happened almost seamlessly. It's because both like CNCF. There's always somebody, someone ready to make what is now private open sourced and that's the great thing about the community. You spoke about what do you think about infrastructure as a code standardization? I go back to open standards, right? Now only if your infrastructure is written to a standard specification and in the old days, they called it terraform providers. I'm calling it old days as though it was decades ago. It was just a year ago, but.

>> In tech land, a year is a decade so.

>> Dog years they were saying in Kubernetes, right? But that standardization allowed us to write code for the infrastructure. If I had gone to someone 10 years ago and said, "You can deploy on an Intel machine the same way you deploy on an AMD, the same way you deploy on a Mac," they'd go, "What, really? You can actually do it." And now people don't even think about it. A lot of my customers are what we call shape agnostic. They don't know what hardware is running. They just know that containers are running to the scale that they want to and scale back down to zero. They don't care what infrastructure you run on. As long as it's cost efficient, really nobody cares and that is what standardization brings to the world. And not just that making infrastructure so ubiquitous means the developers, most of who are trying to bring value to their customers, don't have to worry about infrastructure. It's there, it's always there. It is standard, it's scales on demand. What more can the developers ask for? So standardization on infrastructure as code is definitely coming and we are at the forefront of doing that. Not just for the traditional cloud workloads. We're trying to do it for the AI workloads too. Acute flow is a great example where started democratizing what data scientists do which wasn't a thing again, a few years ago. But now everybody uses it.

>> Right, and bringing it to the platform engineering aspect of it as well, so yeah.

>> Which has definitely been a theme. So you're at Oracle, you're on the CNCF board, and you are also a part of the Oracle Women's Leadership Seattle Chapter, founding member, correct, which is super awesome. Do you sleep is my initial question here because you wear many hats?

>> I try to.

>> Nap occasionally, periodically when the moment strikes. So how are you and Oracle helping grow our community of women in this space 'cause we're still pretty outnumbered?

>> Yes, so what's been great about coming to conferences like this? First thing that when I came to this conference, Wednesday morning, 8:00 AM, there was a woman's reception that happened. I'm sorry, I forgot the sponsors, but I met a woman who was a PhD in physics, physics, and she was here to learn Kubernetes, fascinating.

>> I love that.

>> I met people from marketing, from sales, from development, so women are everywhere. It's just we don't seem to grow the women from the bottom of the stack, the funnel coming in. That was a problem. We've gotten better at it. It's not great, but we've gotten better at it. Now we need to focus on the middle tier to get them a little bit higher and primarily I try to do that with a lot of mentoring. I am a huge fan of mentoring as a concept. I am not so great at networking, but mentoring, I love. I have benefited from being a mentee to so many different men, women, no difference. But that has propelled me in directions that I would've not thought of without them sort of being there as friends and guides to take me to that next step. So I like to give that back to the community and I do a lot of mentoring myself. Here I met some wonderful women. Again, the Priyanka who did the keynote.

>> Yes, yeah, we had her on the show on Wednesday, yeah.

>> Yeah, and you can see CNCF board, not just the board, the CNCF Foundation is only about 50 people but they are over 30% women. And that's great, they built that on purpose. I was talking some other gentleman yesterday and he was saying we don't need to do this as women and men. What if we take the gender out of this equation? Just evaluate us for who we are, what experience we bring in, and let's go from that basic sort of neutral boundary. and then we will see women shine as much as men do. And maybe more, I wear a lot of hats.

>> Yeah, yeah, yeah, well, I think what you're bringing up is really important and it's not just about that gateway. It's about elevating women into leadership and making sure that they feel empowered to take that step. What would be your advice to a woman who might be watching this right now who knows nothing about Kubernetes like the woman you met who has a PhD in physics? What would your advice be to that person about entering the space?

>> So take the trainings, right? Everything today, in this world where data is in our fingertips, we get a lot of data. But it's my true belief that when you take that first action from the data that you get, no matter where you hear it from, that's when it really sticks and what's the worst that can happen? You can say this is not for me, but now it's not that nagging little thing in the back of your mind saying, "I should have tried it, I should have tried it, I should have tried it." So take the action, go to CNCF, look at the training materials. There's so many free trainings out there. Try to set up your own Kubernetes cluster. If nothing else, really I don't know how many people watched the keynote this morning. Some of the biggest contributors now to Kubernetes, they started their first contribution was changing a line of comment. Somebody said their first contribution was changing a legal contract. You don't expect them from those kind of comments, first comments into Kubernetes from people who are very well known in the community, but that's where they started. So do something about that stack.

>> Start small.

>> Absolutely.

>> I think that's really great. All right, last question for you 'cause this has been a fascinating interview. What do you hope that you can say next time you're sitting here on the show with us that you can't say today? So that could be in Salt Lake or London. What do you hope happens in the next six to 12 months?

>> The acceleration of AI-driven development for the betterment of humankind.

>> Rob: Love that. I just got goosebumps, yep.

>> I am a huge proponent of climate change. I really truly believe that if we don't take action today, tomorrow is going to be really hard. You live on the west coast of United States, so do I. We've seen the number of wildfires every year grow up and up and up, and it's real. It's affecting us, everyday lives, and if you don't do something today, it's never going to get done. So I hope six months from now I come back and I see AI driving that change.

>> Well, I hope we can have that exact conversation 'cause I'm here for it in Salt Lake City when we're at KubeCon there. That would be fantastic.

>> I'll be there too.

>> Oh, see, so we're going to do it. Hopefully we're saving the world at that point. Sudha, thank you so much for being on the show. You're absolutely fantastic, Rob, a pleasure as always. And thank all of you for tuning in for our live coverage here at KubeCon, CloudNativeCon in Fantastic Paris. My name's Savannah Peterson, you're watching theCUBE, the leading source for enterprise tech news. (bright music)