Clips
News
More from CDOIQ Symposium 2024

Shayde Christian

Chief Data & Analytics Officer

Cloudera

Three insights you might have missed from theCUBE’s coverage of the CDOIQ Symposium

The artificial intelligence revolution is changing industry understandings of what a chief data officer is. AI infrastructure demands have driven the evolution of the role beyond the merging of the chief data and analytics officer positions.Unpacking the role of the CDO in the age of AI infrastructure demands was a central focus of last week’s CDOIQ Symposium. Things are getting increasingly messy, according to theCUBE Research Chief Analyst Dave Vellante (pictured).At CDOIQ Symposium, theCUBE set up for full coverage to discuss AI infrastructure demands and more

Cloudera’s true hybrid model provides foundation for building AI use cases

With more than 25 exabytes of data under management, Cloudera Inc. has staked out its position as a central player in the delivery of enterprise artificial intelligence, automation and advanced analytics. The company’s approach is what it calls a true hybrid model, a solution that provides the foundation for building and deploying AI applications.TheCUBE on set at CDOIQ 2024.“True hybrid … is where you have harmony in a multiplicity of environments, where the environments are working cohesively together to allow you to democratize data effectively and to generate insights,” said Shayde Christian (pictured), chief data and analytics officer at Cloudera

play_circle_outline Role of chief data and analytics officer at Cloudera.

play_circle_outline Evolution of Cloudera beyond Hadoop to multi-cloud and AI accelerators.

play_circle_outline Importance of common data and governance in modern data architecture.

play_circle_outline Integrated stack as a moat for technology companies.

Info
Transcript

Shayde Christian | Cloudera

Shayde Christian

Chief Data & Analytics Officer Cloudera

Dave Vellante and Sanjeev Mohan interviewed Shayde Christian, the chief data and analytics officer at Cloudera. Shayde discussed his role in delivering services to the company's executive staff and overseeing AI-related services internally. He highlighted Cloudera's evolution beyond being a Hadoop company and emphasized the importance of reliable and responsible data for AI initiatives. The conversation touched on the challenges of implementing gen AI effectively and the value of bringing AI to the data rather than moving data for energy requirements. Shayde ... Read more

explore Keep Exploring

What is the scope of the role of the person in charge of data and analytics internally, including AI aspects? add

What are Cloudera's superpowers when it comes to delivering business value and ensuring trust in AI through reliable and responsible data? add

What strategies are being implemented to harmonize and integrate different data environments and models in a true hybrid approach? add

What is the biggest advantage of companies like Cloudera and hyperscalers in the technology sector? add

bolt Powered by CUBE AI

Shayde Christian | Cloudera

>> Hi, everybody. Welcome back to CDOIQ from Cambridge, Massachusetts. My name is Dave Vellante. I'm here with Sanjeev Mohan. My cohost Paul Gillin will be in house tomorrow. This is our eighth CDOIQ, the 18th event. We've done well over 100 interviews. We've got 150 guests. We're really excited to have Shayde Christian, who's the chief data and analytics officer at Cloudera. Shayde, thanks for coming on with me and Sanjeev.

Shayde Christian

>> I'm privileged to be here. Am I your eighth or 18th interview of the day?

>> I think seventh maybe. - Seventh. We're getting there.

Shayde Christian

>> We've got a ways to go now to match the body of work

>> that we've created here. Chief data and analytics officer. I'm waiting for AI to get into that title. We've been talking all day about those CDO and the AI roles coming together, but talk a little bit about your role, the scope of that role.

Shayde Christian

>> Yeah. I'm in charge of data and analytics internally, so I deliver services to the executive staff in their organizations. I'm also in charge of AI as it relates to that service delivery internally. We're a technological company, and so we have different AI factions within our product development headed by Dipto, our CPO.

>> Ah, okay. You're a practitioner?

Shayde Christian

>> In part, and then because I utilize our products and services and my team utilizes them, there's an external facing function here to evangelize and talk to other CDOs about how to do this.

>> I'm sure the sales team wants to drag you all over the world.

Shayde Christian

>> Marketing and everyone.

>> Best practice. theCUBE actually started in Cloudera's offices. Did you know that?

>> Oh, no way. Where? - Oh, my God.

>> Right next to the old Fries. You remember that?

Shayde Christian

>> Yeah. - electronics.

>> It was next to box.

>> John Furrier put out a tweet saying-

>> In Palo Alto.

>> Yeah. In Palo Alto.

>> Saying, "Hey, I'm trying to find a place

Shayde Christian

>> to cohabitate . " I said, "Well, why don't you hang out at Cloudera? " We did a quid pro quo. We said, "We'll create a cube studio inside of Cloudera and exchange for the space. " We did. That's how theCUBE got started, and then we had one on the East Coast, too.

Shayde Christian

>> Am I part of the exchange? You're still paying this off?

>> No. You guys kicked us out when you needed more space, but point is back then Cloudera started the big data movement. You were associated at a time with Hadoop. You hired Doug Cutting, who was the founder of Hadoop. But you've evolved dramatically since that. Obviously, merged with Hortonworks. What's the update on Cloudera? What are you up to?

Shayde Christian

>> Yeah. We're no longer a Hadoop company. That's one block in our huge pyramid of innovation. We've had a lot of innovation since, as Sanjeev knows, where analytics at the edge, multi-cloud. We have accelerators for AI that let you bootstrap these foundational models very quickly, but none of it really matters. Those are just blocks in our stone. It really matters what we're delivering in terms of business value. To trust your AI, you have to trust your data. Your data has to be reliable, has to be responsible, and these are Cloudera's superpowers from the time theCUBE began until now.

>> Sanjeev, you put a link into the notes here on this Goldman report, which essentially, I'll just read the title. You can search. It's a report that's kind of negative on AI generally, but it's gen AI specific. Gen AI, too much spend, too little benefit. I don't know. It's maybe a little controversial. Maybe it gets some good eyeballs, but I'm not surprised. We've just wrote as well, it's still elusive, that ROI. Takes time. It's not some magic technology. But what are you seeing with customers in terms of how they're applying gen AI, and what are you advising them?

Shayde Christian

>> AI in general, our customers have been employing AI in Cloudera for some time and are achieving massive amounts of business value. This was a little bit more specific on gen AI, and I understand the point. If you need to develop a landscape of Nvidia architecture to build your own LLMs to do A, B, and C, then you're probably going to overspend and under deliver.

>> Big denominator. - But there are so many ways

Shayde Christian

>> to employ gen AI cost effectively and for good use cases.

>> Just even a small one. I was part of watching a group. One developer for two months, 99. 99% automation of contract summarization and managing contract clauses, and you do that by bringing the models to the data. You bootstrap them where you can train them internally, and that group trained them on CPUs. There are very cost-effective ways to do this. Foundational models are becoming commodities, just like your compute and storage. At any phase of these kind of revolutions, of course, you're going to overspend and under deliver. But like any technology, this is going to come together.

>> I also want to make a point. You're bringing AI to data, so having data under your management is really important. One thing that Cloudera does not talk enough about is that Cloudera, actually even today, has one of the highest amount of data under its management, which is 25 exabytes. It's actually far more than many of the big companies that we talk about on the show.

Shayde Christian

>> That's 25% of the world's data. That's enormous. Yeah. We're not good at being-

>> Wait, Cloudera software touches that 25%?

Shayde Christian

>> Our platforms host 25% of the world's data. Massive.

>> Irrespective of physical location, obviously.

>> Correct. - That's interesting.

>> As analysts, we hear a lot the repetition of things like,

Shayde Christian

>> "We're going to bring AI to the data."

Shayde Christian

>> Correct.

>> We've heard Jensen say that at a variety of conferences. We've said it. I've said it. Howie Xu. I don't know if you know Howie. He's an AI guy, former security companies, Palo Alto and Zscaler. He came out with a contrarian view. He said the energy requirement is going to be so high.

Shayde Christian

>> That is true. - You're actually going to have

>> to bring the data into where the energy is, and that's going to be in the cloud. Now, that's contrary. It's very contrarian. We produce the gen AI power law. We talk about it all the time. It's a power law, but it basically says, like a lot of power laws, you're going to have a long tail across the enterprise. The vertical axis is model size, the horizontal axis is domain specificity. You're going to have a lot of on-prem activity. You guys are multi-cloud. You don't care whether it's in the cloud or on-prem, but what are you seeing in terms of domain specific AI generally, but specifically gen AI in terms of being adopted where the data lives? Are people moving data into where there's energy?

>> Energy. Yeah. - Because people hate to move data.

>> Can I say something? I completely agree.

Shayde Christian

>> The energy requirements of AI are astronomical

>> and we know that, but then if you read the Economist, according to the Economist, the energy problem of the entire world is soon to be resolved, and that is through solar.

>> That's what Elon says.

Shayde Christian

>> Actually, what's happening is China, in order to get out of the economic malaise they're in, they've oversupplied. They have overabundant factories that they're built, so they're literally dumping solar cells, which are very efficient these days, all over the world. That's their way to get-

>> Ray Kurzweil says the same thing, by the way. Exponential growth. He says, "If you do the exponential curve," he goes, "This problem will be solved."

>> Correct.

>> That's interesting. Anyway.

>> Given that you have a broad observation space across all that data and all those customers, are you seeing the domain specific AI pop up and produce results?

Shayde Christian

>> For us, what we look at is true hybrid. Hybrid to me is something where you have data and apps in multiple locations. True hybrid, I want to differentiate. It's somewhere where you have a harmony in your multiplicity of environments where the environments are working cohesively together to allow you to democratize data effectively and to generate insights. We're the only true hybrid company in my opinion. The tenets of true hybrid really are about interoperable, portable data services. I want to build a workload one time and I want to run it anywhere in the world on any platform, and it works instead of having to rewrite it in AWS and Azure on premises. Open data lake houses are a big tenet of true hybrid. Having a central command and control for security and governance is very important. To me, a practitioner having the full data lifecycle of tools available to you in integrated fashion is another tenet of true hybrid. Yes, I agree with you. We're already doing a great job at Cloudera of bringing the models to the data, but you've got to take the data to the models as well. We're looking at that true hybrid harmony of multiplicative environments working together as a way to accomplish it.

>> How do you harmonize that? Is it an abstraction layer that you apply across those different estates and different models, or how do you make that opaque to the customer? I mean that in a good way.

Shayde Christian

>> Common form factor. You're on one form factor. It doesn't matter what the form factor is. You have common code is one thing. Another thing is you don't have to rewrite software. We all had a lot of pain when we migrated to the cloud, because we had on-premise software, and now we had to acclimate to our AWS store or Azure store. We're doing it at the level of software and form factor. Common code for form factor and same code. You write it once, you run it anywhere, and there's no amount of rewriting. Optimization is always going to have to happen. If you're going to go to the cloud and save on compute and storage, you got to optimize your workloads.

>> One of the ways that Cloudera has done it, from what I've seen is, that they very quickly got into S3 compatible storage , and now with Iceberg. Actually, Cloudera adopted Iceberg way before it was even a thing. That's how they're standardizing, how they're writing applications. Then after that, what they do is they take these applications, they dockerize it into a container, and then they can run it on prem so you don't have to rewrite it. That's how you get this hybrid benefit.

Shayde Christian

>> Love it. - What's your take on the whole trend toward

>> OpenTable formats? Everybody wants now to be able to control their own data, bring any engine. May the best engine win is the mantra now. How do you think about that? How are customers thinking about that? How do you think about it as a practitioner? How do you think about governing that open data? Sounds like a free for all. How do you balance the need to reduce lock-in, control your own data with it's got to be governed and I got to generate business value?

Shayde Christian

>> We're highly against vendor lock-in. That's why we're almost purely open source. Our SDX components governance and security have some wrappers that we provide as well, but our offerings don't even lock you into us as a vendor. Open is still very big with us.

>> What about the governance aspect of that? How will customers solve the governance problem while at the same time preserving that openness? How do you think about that?

Shayde Christian

>> I don't think I can sugarcoat this. I've been surprised. I've been attending a lot of conferences since 2021, and before gen AI exploded, I'm still being asked the same questions from stage and panels, which is what is your enterprise data strategy, and how do you do data governance well? I think what's wonderful about our platforms is that we provide common data and governance. Same way to do it on any form factor, but it's just the starting point. Governance is really, really hard work and we can make that uniform, but nobody can really make that simple.

>> How should we think about where you add value? I'm talking about Cloudera as a vendor now, not so much, Shayde, as a practitioner, but I'm interested in that as well. Oracle would say it's in the DBMS. They're not relinquishing that value. We've talked about, and I don't know if you buy this, but it seems to me that the DBMS in terms of the source of control, it seems to be moving to the data catalog. But that's not where the value is. It's all being open sourced.

>> Correct. - Where is the value

>> equation now?

Shayde Christian

>> Is it the ability to harmonize the different data that's out there? Is it the ability to work multiple tool chains at the same time? How do you see that?

Shayde Christian

>> Data is everywhere. How do you break down the silos? How do you get it all connected in together with a common set of tools to govern, secure, manage, process, collect, and analyze it? To me, that is the common element, getting commonality and harmony with one place to do everything, and you can take it anywhere in the world you want to go with it, including this discussion we've had about bringing models and data to each other. I still think this true hybrid, and part of that true hybrid really is modern data architecture. It's about data fabric and data mesh. I just still think that's it. I hate to give you the same answer twice, Dave. I feel you deserve a second.

>> It's all good.

>> Dave, I have a point of view on this. If you look at what's going on in our industry with these rapid changes that are going on with new standards and new ways of doing things, there is really very few moats that are left. Technology is not a moat anymore, because if you're working on Iceberg and catalogs are open source, then what is your moat? There are only a few moats. One moat is capital and skills, because OpenAI and all these companies.

>> Cloud vendors have the balance sheets.

>> Yeah. Cloud vendors have the money. But the biggest moat

Shayde Christian

>> from technology sector in which Cloudera does really well, and so do hyperscalers, is an integrated stack. The companies that have an integrated stack that brings together structured and unstructured data. Also, batch data, streaming data with any compute engine on top. That compute engine could be Spark, it could be Pandas, it could be SQL, DuckDB, or it could be an LLM and you could do RAG and vector embeddings. Then on top of that, you build your data products or AI products. That to me is the moat.

>> Interesting. It's one thing to have the tools, because I would argue that AWS has the tools, but that stack is not integrated. It's just their tools in the toolbox that you have to then integrate.

Shayde Christian

>> Correct. - If you're going to be multi-cloud,

Shayde Christian

>> they certainly don't integrate with Azure, and that's what I meant about this harmony and the true hybrid nature of it. You want one place to do everything no matter where your data is. What you're saying is exactly true, but once I get to that level, I want to be able to do it once and do it everywhere. To me, that's the secret.

Shayde Christian

>> You have unified governance irrespective of where the data resides and what format it's in. That unification and integration, in fact, the pendulum has shifted. It used to be modern data stack where we had hundreds and thousands of products that we were dealing with. The pendulum has now shifted to a more centralized way, because what businesses are saying is that, "We don't have the time or the budget to stitch together different things. We want the vendor to provide us a data fabric or some prebuilt, pre-integrated components."

>> We rewrote our applications to go on the cloud

Shayde Christian

>> and then we learned it was more expensive there. Now, we're repatriating. If you're not on that common data platform with common tools, that's so much energy wasted and redoing. I want to do it one time. I want to do it one time.

>> The other bromide that we hear is you got to get your data house in order before you can do AI. It makes sense, but how much of your data house? If you have to get your entire data estate in order, you'll never finish painting the Golden Gate Bridge. How do you think about that as a practitioner and as an advisor to your customers? Can they get part of their data house in order to solve a specific business problem, or do they actually have to at least have a strategy to get the entire data estate in order before doing it?

>> I like how you ended that sentence.

Shayde Christian

>> I like the strategy forward, but yes, it all depends on the use case. There's an insurer who looks at training AI models to recognize from satellite images how devastated an insured home is after a natural disaster. Partially devastated, somewhat devastated, fully devastated. The AI models offer that to an individual, say, "Do you agree? " An ACH payment is cut the next day to repair the cost of that home or a new home. For that, you need external data. You don't need to aggregate all of your data. But there's a biopharma company who's trying to significantly cut down the duration of time it takes to get lifesaving cures into the marketplace. In that super complex, hugely valuable use case, you need a lot of data to be aggregated. It really depends on what you're trying to accomplish, but there's a lot of low hanging fruit.

>> All right. We got to leave it there, Shayde. Thanks so much. theCUBE.

>> Appreciate you. Thank you. - It's good to have you.

>> Thank you you so much for coming.

>> Thank you. All right. Keep it right there, everybody.

>> We'll be right back. Dave Vellante, Sanjeev Mohan.

Shayde Christian

>> You're watching theCUBE's coverage of CDOIQ from Cambridge, Massachusetts. The 18th annual CDOIQ. Be right back.