Google Cloud's Evren Eryurek - Finding the business value for moving data to the cloud

Evren Eryurek:

You have great set of talent in every company and they're all, "Look, oh, I'm going to hire a 100 data scientists."

"Why?"

"Well, they're going to do this for me." "Well, you may actually surprise yourself. You have great talents, enable them to ask those high value questions, enable them to get really access to this data that you are thinking you have and get something out of it. That is priceless for everyone."

Anthony Deighton:

Welcome to Data Masters.

We have a great guest for you this week, Evren Eryurek of Google Cloud. Evren is the leader of data analytics and data management portfolio of Google Cloud covering streaming analytics, data flow, B Messaging, data governance, data catalog and discovery, and the data marketplace. He joined Google Cloud as the Technical Director in the CTO office of Google Cloud, leading Google Cloud and its efforts towards industrial enterprise solutions. Prior to joining Google, he was the SVP and Software Chief Technology Officer for GE Healthcare. Evren began his GE career at GE Transportation, where he served as the General Manager of the software and solutions business. Evren is a graduate of the University of Tennessee and holds a master's and doctorate degree in nuclear engineering. And he holds over 60 patents and is the author of a book titled Data Governance: The Definitive Guide.

All right, well welcome Evren to Data Masters. It's a pleasure to have you. Maybe we could start the conversation more on a personal level. I'd love to, if you wouldn't mind share a little bit, you have a really unique career background where you worked and a career trajectory that I'm sure our listeners will find interesting. So maybe you share a little bit about how you found yourself working at Google.

Evren Eryurek:

Really great pleasure being here [inaudible 00:02:12] with you and your listeners, it's so exciting and thank you for the time. Yeah, I have been very fortunate in my career to work with some great companies. I started right after my PhD in the process control industry, where we literally revolutionized and we went from analog to digital. That was the reason I was hired into it and I led the technology teams there. And so it was sort of what I have done in my PhD years, we ended up doing it in the real world. You know, how could you say no to that? And nowadays we call it IOT, believe it or not. Now we were able to do all these digital communications, control systems, measurement systems, and now you can imagine what data quality meant for that world. We're dealing with chemical plants, refinery plants, process control industries, very high risk.

And then GE came along and I wanted to run a business. And then I moved into transportation business to run a control systems and monitoring systems for that industry. That was a hoot, I learned a lot. So I applied all my machine critical software and importance of data, importance of control systems. And quickly they pulled me up. They said, "We need you to come and run the healthcare technologies." So, I became Senior VP and CTO of GE Healthcare. That was interesting. So, we have teams everywhere. So you're dealing with the notion of dealing with lives and health just gives you a totally different perspective. You really take everything to the next level in your thinking, how you validate your thinking, how you validate your data, everything matters because you're dealing with health. So, that became a life changer for me and long story short, after some really fun time with great people to work with.

I've been a cloud evangelist and I was pushing cloud and a lot on my friends in the healthcare CIO world, "Hey, Evren we don't want to talk to you about clouds. We will never do clouds." Now, I can see a lot of people smiling after that word, which I was too. And a few years later, they came, "Oh you know what? We actually want to talk about that cloud thing and so forth." And so never say never right? When we launched the first industry, leading health clouds in GE, and I was really interested in Google's data analytics technologies. And I was trying to figure out how we can leverage some of that work because as a platform, we picked another providers for the IS parts. And so, Google changed the conversation and they wanted me to come and join Google Cloud, which was just getting ready in 2015 and the rest is history.

I mean, how do we say no to Google? And I became a cool dad overnight. And I was cool when I was dealing with the choo choo trains and my kids loved it, they have all these locomotives and stuff, even though it had nothing to do with my business. And then I say, "Hey, I'm going to go to Google."

"Really? Wow, so cool." So, just the entire family is now on Google technologies. We use all Pixel and Android so forth. So, here I am and my passion was always data. My passion was always real time, which is what I do, as you know, we've worked with you guys very closely and so the importance of data and high quality data has been in my life since the day one of my career.

Anthony Deighton:

Interesting. And I love this idea that in the work you're doing with GE Healthcare and before, the challenge of whether data can or should live in the cloud, you're really on the forefront of that movement. Where do you see that transition today? So, it's fair to say five, 10 years ago, you were the crazy one. Are you still, I mean, maybe you're still the crazy one, but are you still the crazy one? Or is this cloud migration thing taking off?

Evren Eryurek:

I want to keep my crazy title, that's okay. You know, you got to have some visionaries out there, who got to push it. It's not as crazy anymore. So if you look at... Let's start with the healthcare industry, I mean, it's really important for you too. I am already seeing, if you look at, for example, the genomics world, they have tapped into cloud so accurately so nicely, and they have created this environment where many could collaborate no matter where they are in the world and work on a very high quality set of data that went through their own scrutiny in the highest levels of security they could consider in health because I know how it was before the cloud world. You know, hey, you probably saw my book, I have an example in it from the healthcare industry, how data governance is important, but you know, when things get hacked, it does happen in every industry.

So, the healthcare industry is moving some slowly, but different parts of the segments of the healthcare industry is moving faster. And I don't think we can prevent the progress in technology, it will happen. So people are more used to it. They're no longer asking or saying, "No, we're not going to do it, this is stupid." They're saying, "Well, how do we do it? Where do I start?" Every day, we make an announcement just this week we announced another big healthcare player partnering with us to move to cloud. I've been working with them, I know them for many, many years and it's a matter of time globally to really get to a place where you actually deliver better care and much faster easier solutions and maybe personalized care in many cases, because now we're enabling them. Look at what happened with COVID. Every one of those industries tried to figure out how they can collaborate with them. I think healthcare teamed in their best. They all already have a lot of these technologies in hand, they were able to accelerate whatever the solution they were looking for, it all happened because of these technologies.

Anthony Deighton:

So, I think one of the important ideas embedded in what you're saying is that there was always a business value or business context. So, it's not about moving data to the cloud for the sake of moving data to the cloud. Why? What are you getting from that? Is it increased collaboration, better security, the ability to accelerate outcomes, understand it analytically? Is that a fair way of thinking about it, so the business value first?

Evren Eryurek:

Absolutely, think the business value. People come to us, look I've talked to so many CIO, CBOs, you name them, they [inaudible 00:09:57], "We want to do analytics." That's awesome. What kind of analytics, what are you trying to get out? What is the high value question you're trying to ask and seek an answer for? And then some are prepared, they [inaudible 00:10:12] "What we want to do." But most are after some kind of a buzz. So, if they're going to do it in the sake of doing it, [inaudible 00:10:21] let's think about it. And then they go, "We want to do big data analytics." Now, all they did is add big data to the word analytics that didn't change anything. Then of course the famous, "Oh, we want to do AI." But if you're not ready to deal with data just for the sake of moving data, don't move it.

We really have to understand what data means, how these disconnect, it seemingly disconnected set of data that might be available out there. How do I make sense from it? How do I get some insight from it? Data is out there, but is it actionable? That's what the value is. You have to start from it, no matter what the industry is. The first question is, "Can I act on that data? Do I have the means to it? Can I collaborate? Can I bring the right eyes to it?" And I always believed in, if you bring in right folks and enable them in a governed secured manner, they will get insight from it.

Anthony Deighton:

So, I think it's a really important idea starting with the business value. But one of the common refrains we've heard from customers is that their data is a mess, the data is a disaster. And I think it'd be fair to say that moving your data to the cloud won't improve the quality of it. And I've used this analogy in the past, you know of moving house. And-

Evren Eryurek:

[inaudible 00:11:56].

Anthony Deighton:

... you don't move a messy house, you clean your house and then you move. And you know, it feels like the same is true as it relates to data moving to the cloud. But yeah so what's your view? I mean, do you move first and then clean or clean and then move?

Evren Eryurek:

Look, that's why it's a journey. And the way that I have been approaching is, if we have the means to do, sort of the housecleaning, like you said ahead of time together, I think it helps. It helps you from multiple ways. Now, in some cases they may not have that option. They may just be in a hurry, "You know what, we're going to do this thing." I get those. But if you are a long journey, that's why we believe in this hybrid nature of the world that we are in. It's going to be hybrid. There is no flip the coin and then switch from what you have today to cloud. That's a journey beginning of that journey should start thinking about what am I going to do with my data? What do I need to do with my data? Do I know what I have? A lot of the folks don't even know what they have, where they have.

And in some cases they have the same set of data in four or five different ways. No one knows what's the mess and what's not. So doing some of that housekeeping, if they have the opportunity ahead of time, definitely helps. And they have tools like, [inaudible 00:13:29] is a correct tool in here. We have tools already, capabilities already in place to help accelerate the journey. And then what happens is, or the accelerated thinking. "Oh, I brought my data. I want to see the value out of it now."

"Well, you know you just brought a big mess, you weren't thinking ahead." We really have to think about how we can get to the insights, how we can get the value fast.

Anthony Deighton:

So that's great. And I think it feels like one of the primary inhibitors to moving data to the cloud, is that it's trapped in silos. That the natural behavior, when I have data on premise or in my environment is that I store it, by division or by product line or in a different data infrastructure, different databases, and moving data to the cloud becomes this moment in time to break those organizational or product or technology data silos, and really unlock a lot of the value you're talking about.

Evren Eryurek:

Yeah. That's why I keep highlighting one of the biggest things cloud did for many industries is improving the collaboration. Cross-collaboration, industry-wide collaboration is an awesome element to it, but collaboration within the company-

Anthony Deighton:

Yeah, we can start inside the walls.

Evren Eryurek:

... Priceless. Let's solve that together. You have great set of talent in every company and they're all, "Look, oh, I'm going to hire a 100 data scientists."

"Why?"

"Well, they're going to do this for me."

"Well, you may actually surprise yourself. You have great talents, enable them to ask those high value questions, enable them to get really access data that you are thinking you have and get something out of it, that is priceless for every company."

Anthony Deighton:

So you'd mentioned this before, I just want to pick up on this, the idea of breaking those data silos, using Tamr and specifically kind of using a machine learning-based probabilistic approach to breaking data silos, gathering together the data, improving the quality instead of a manual rules-based coding first type approach, a traditional approach to breaking those data silos. I'm curious on your perspective as a sort of a cloud expert, why is this machine learning-based approach better implemented natively on the cloud or built on the cloud?

Evren Eryurek:

Okay. Remember, you're talking with a guy who did his master's and PhD in the world of AI before we had all these AI platforms. And I actually spent a good part of my graduate studies on so-called expert systems remembered those rule-based engines and so forth.

Anthony Deighton:

Sure.

Evren Eryurek:

Eventually you get to a place and that eventuality is not really millions of rules, it's very finite small number of rules, you start conflicting. You can't rule your way out in this complexity. And the rule of thumb is if a machine can do it and this is true in every industry, if a machine can do it, let it do, that is the rule of thumb. And machines, machine learning, and discovering these patterns, achieving these they're really good at it. And that technology will improve, they brought it into place where it's in our lives every day.

So creating rules, there are going to be some rules, don't get me wrong. There are going to be some augmentation that we're going to do with it, but bulk codes can be achieved, 80, 90% of it can be achieved, machine learning, and it eases everyone's minds. Now we have to be able to validate it. Now I'm going to put my healthcare hat on. I have to be able to validate how I did, is repeatable with this machine learning approach. That is an important aspect, but the tools do it today. So, it accelerates that kind of the grungy word in the beginning that we have to do that, nobody wants to deal with it. And you'll no longer trapped in the minds of a few people who may actually understand the entire rule-based system. You're actually creating an institutionalized approach in my opinion, that's one of the reasons why I like machine learning parts.

Anthony Deighton:

So it's really about leveraging machine learning, but the human guidance and feedback is critical. And then I would add, I think a perspective which is on the cloud, or maybe I'll say this a little differently, these machine learning algorithms are incredibly compute-intensive. And so if you can take advantage of cloud scale compute, that allows you to maybe more practically running and execute this machine learning in a way which may not have been possible on [inaudible 00:18:57].

Evren Eryurek:

Very much so. Take a look at my world. We're dealing with many of our customers, hundreds of thousands in many of them, millions of events per seconds, we process. And this happens at a minuscule level of compute because we provide so much compute in such a short period of time. We can actually [inaudible 00:19:21]. Imagine doing this thing in your old worlds. This is just not doable, it's going to cost so much. It's going to be so hard to do it. Now, this is not in the fingertips of everybody. You take your products, many with capabilities like Dataflow, BigQuery whatever, you have a great environment to accelerate this and do it in a fast way. And not only would you be starting to clean up your data and bring it on and start finding insights, but you're taking advantage of this advanced technologies from get go, right away and get the value out of it. It's all in there for you.

Anthony Deighton:

So, yeah, so we're bringing these ideas together. Not only is it a good idea from a business value, security, analytic perspective to get data into Google Cloud or into the cloud, but now it opens up the possibility of, you know it becomes a forcing function for cleaning the data and gives you better tooling for doing it because you can now take this machine learning-based approach, leverage Tamr to get that done. So, I want to shift the conversation a little bit to Google. So, we've talked a lot so far about these sort of like generically it's a good idea to move data to the cloud. You know, generically, it's a good idea to clean it. Yes, probably machine learning is the best way to do that. You know, it strikes me that Google is really on the forefront of a lot of innovation and cloud technology. And I'd love you to take a moment and speak to the investments Google's making and some of the unique capabilities. You know why should a listener move data to Google Cloud and not somewhere else?

Evren Eryurek:

Wow, thank you for giving me that opportunity. This is, I don't know if your listeners got a chance to see it, but yesterday we had a massive event Data Cloud Summit was there and we announced whole bunch of new capabilities. It is all in the interest of making it very easy for the industries to take advantage of the massive assets, the data assets that they have. It's all about that for us. You know talking about from bringing the data in ingestion, processing the data with the capabilities like Dataflow and BigQuery and so forth. And generating insights with our Looker and BigQuery and Dataplex, we made a lot of new product announcements there. It is all for bringing data and getting insights and enabling the ecosystem. We mentioned Dataplex yesterday, which is our lake platform. It is primarily there to create this collaboration for the companies that you're seeking.

We're trying to think how we can make this journey easier for them. And I've been always pushing to improve what we can do in the data processing world. I actually had a blog about it. We talked about how we're changing the data processing world in Dataflow time announcements. Where we're doing the entire infrastructure monitoring and sizing for the customers. So that we're taking all these complexities away from them, so that they can focus on their value add. Guess what, they are all machine learning driven. It is the bread and butter of it. We have it in there, we can actually optimize it much better to do it that way and much more efficient to do it that way. For us, it's all about, "Can I make it easier, the journey? Can I accelerate their journey? Can I get into the insights of what they're seeking, so they can get value faster?" That's all for us.

Anthony Deighton:

I mean, that's amazing. And maybe if you could cast your eye into the future and maybe to kind of connect where we started this conversation, your career trajectory, what brought you to Google, if you had advice for maybe even a young listener, somebody who's early in their career, you know, they think what Google Cloud is doing is interesting, they sort of agree generally the data should move to the cloud, or you should probably clean it up using a machine learning-based approach. They've bought into everything hook, line, and sinker which would be great. What advice would you give them for the next five years of their career, as it relates to data, insights from data, improving data quality, leveraging the cloud, and how do they take advantage of that? How do they lead into that from a career perspective?

Evren Eryurek:

Wow, that's a good one. I was talking to set of graduate students just yesterday in a very big Yale University. I think these are great times for being students or being early in the careers. First of all, I would encourage all of them to get really into whatever the Coursera that they can pick, courses that they can pick, get in it, get their hands dirty and get your certification. It will pay dividends guys and be the evangelist within your team. What we're seeing is, in each of these companies that we're working with, there's always a generation of evangelists, that they're trying to change the way that they're doing it, the way that they are actually thinking about it. This digital transformation, quote, unquote is happening because of this generation. They're seeing it. Now, if I look ahead, these people in the next five years who are in the front end of their career will be the ones driving this because they are so comfortable.

I tell you, my kids are so comfortable with doing things that you know, you and I wouldn't do it when we were young. We wouldn't touch those things, [inaudible 00:25:44] just do this and I'm going to pull it together. That comfort is hugely important for every company. Give them back, be the agents, take it, "Okay you know, I'm in this. I'm actually learning data processing. I'm learning, getting insights out of it." These are free, go get those credits, start playing with it, put together a proof of concept. Guess what? Everyone will latch on it, they will love it. You will shine, be the spear of the [inaudible 00:26:15].

Anthony Deighton:

Yeah. And to your point, there's both a lot of free resources, but also even the paid resources, but like the amount of technology that's available at the fingertips of anyone that wants to take advantage of it, you know it's phenomenal also the academic research. I mean, even going back to looking at kind of the core algorithms behind Tamr, around that machine learning, you know a lot of that is available in Google scholar. You can look up the academic papers and the patents, and then you can leverage Google Cloud technology to try it out.

Evren Eryurek:

Yeah.

Anthony Deighton:

[inaudible 00:26:55].

Evren Eryurek:

Try it out. There's this fear in some, "Oh, how am I going to start? How am I going to do this thing and so forth?" You can't go wrong. I mean, in the [inaudible 00:27:07], you can delete and start from scratch. That's okay. That's simple in the world of cloud.

Anthony Deighton:

Well, Evren, it's been a real pleasure and certainly appreciate the partnership with Google and with you personally. And I really appreciate you making the time and the insights that you've shared with our listeners. So thank you.

Evren Eryurek:

Thank you for giving the opportunity, the pleasure is mine. You guys are awesome partners and deliver huge value to our customers. Really talking to cloud native data quality providers, it's look, you know it's dream come true for me. Thank you. Thank you for the partnership.