A Roadmap to Data Mesh Success w/ Omar Khawaja

Media Thumbnail
  • 0.5
  • 1
  • 1.25
  • 1.5
  • 1.75
  • 2
This is a podcast episode titled, A Roadmap to Data Mesh Success w/ Omar Khawaja. The summary for this episode is: <p>Some of the greatest breakthroughs throughout human history can be attributed to the biotech field, from life-saving vaccines to advances in supply chains. So it’s no surprise that biotech is also a fast-mover when it comes to innovative socio-technical paradigms like data mesh.</p> <p>In this episode, Tim, Juan, and Omar Khawaja, Global Head of BI at Roche, break down the good, the bad, and the ugly of the company's data mesh journey.</p> <p>This episode will also feature:</p> <ul> <li>How to identify potential use cases and pick the one to prioritize</li> <li>How to think about ownership when data and tools are widely distributed</li> <li>You’ll be at Snowflake Summit in Las Vegas next month. So will we. What’s the best buffet food?</li> </ul>
Warning: This transcript was created using AI and will contain several inaccuracies.

This is catalog and Cocktails world.

Hello. Hello. Hello everyone. Welcome to catalog and Cocktails presented by the Enterprise data catalog. For the modern-day to stack. We're coming to you live from Austin, Texas somewhere else. You'll find out in a minute. It's an honest. No BS, non-sales. A conversation about Enterprise data management, with a tasty beverage in Han Kim. Gasper long time. Getting hurt and Prada guy. Did it Out World joined by one day Tim. I'm want the kid on principal scientist a digital world and Wednesday middle of the week and towards the end of the day and it's time to talk data today and I'm actually live from Cali Colombia. I think I've been almost every other week somewhere else in the world today. I'm visiting visiting my parents who live here. But anyways today match. Yes, they always comes up. What's today's episode?

Special because we're not going to go do our just blah, blah blah, datamatch. We're going to go really talk about how this stuff is actually implemented the Journey of data mashed and somebody who can really speak to. This is a name that we've been seeing a lot and I've been following her for a long time. Have the opportunity to go talk a lot with him and learn so much, and this is Omar. Khawaja was a global head of the IR troche. I keep couple of week ago or two months ago. He gave this fantastic, talk called the mind-body heart and soul into their data messaging from a nation. And I'm like. That's what I thought when I saw it. I was like, I need to meet Omar and go talk to Omar more. And I'm just so happy that you were here in part of our show today. Omar. How are you doing?

I'm doing fine hun. Thank you so much for having me. Amen, And it's a pleasure to be there. Thank you so much for sticking around here by 11 p.m. I don't know. What, are you drinking right now? So I'm just having my Coke Zero to survive the remainder of the night, and that's it. And the irony is, I'm having in one of my shot glasses collection that I have with you. Maybe seeing at the back of my screen right now. So, that's perfect. And I'm drinking a little bit of tequila and Topo so, very simple today. So that's what I'm sipping on. I'm just having a nice beer.

Club Columbia, very classic beer here in Columbus is really nice Kris lager. And I would have cheered because I'm going to meet you and we went out to go to snowflake Summit and we will be there, you will be there and I'm toasting for more travel and finally got them for cheers to that just for the two years to real life meeting people. Once again, yeah, question following up on that is I hate. You'll be at the snowflake Summit in Vegas next week. So will we what's the best buffet food?

Oh, man.

I think something that you can eat a lot and I am a big fan of eating french fries, and that's the secret of my health. So I would love to go for a french fry Buffet. I have a long rant on what I got a little later, but how about you 10. Now, I'll take some of that. What about you want to do? You have a buffet food. I'm just overwhelmed that. I know I'm going to eat you. I would eat too much and I'm going to gain weight. So I'm just going to go take a bunch of salad and some protein did not look anything else, but here's a funny.

Something about I have never been to Vegas in my life. So this is, there's another more interesting places to visit. What the heck does a date of mystery.

Oh man, it's a maybe others have experienced the same but I got to the chase by the key, right? So I like me many others that it as well. I read it. I ignored it completely. Who is this person? What is she talking about? Because anybody in the space of data analytics, whichever works. You are comfortable with have done everything that she has described in the first part of the article. And then have pointed out the challenges, the issues that everybody fish and the first reaction, allergic reaction to that. And no kidding. I didn't read the article. Once again, it's a very long and I remember back in 2020 when I joined at the computer.

I'm working for right now at rush. I had the opportunity to decide what kind of parts that we are going to take for piety of responsibility. And this topic was quite hot in the community of Architects and data practitioner within the company's and also at the diamond industry and I was like, okay we have done Dejah Lakes. We have done Dejah. How is this even longer? And then we read all the articles of failures of data projects and initiatives and hardly few percentage of succeeding. So what's the cost of that? And sometimes it's technology, sometimes it's people, sometimes the data, sometimes it's processing and it occurred to me that maybe this is the approach, let's explore it. And it was the peak of the pandemic. Nobody was traveling. We had no vaccinations on other things.

Even some tests were not available in an apoplexy and guess what? I just reached out to come back and said, hey you have just turned my life upside down and get the speaker explain. You want to catch up and and she's very kind. She said, yes, let's talk about that. He had three, four, five meetings if I recall correctly and in the middle of those meetings has second article came out. That was just the words, the end of 2020, which had more details. Okay, let's take the next step. What does it look like? Because the first one was quite a while ago, and that's a good to be that clicked with me a lot because the first step of the journey, one was the realization that it's Insanity to do the same things again, and again and expect different results.

And that's a quote from Einstein. And I love it. It's very powerful. And as you might have seen in my thoughts. I was reading this book from Adam Grant, Tinker game, and shout out to anybody who wants to read hits, by all means. Go ahead. You will, you will explode, a wonderful world of how to find new creative ideas, how to challenge the preconceived notions, and those two things, just clicked together to me. And that was the big thing of our journey on how to go about take the first steps for determining, the realization that we want to do that at that time is 2019.

Yeah, this is interesting but I'm missing the me and everybody would start kind of just making stuff up on their own and things. So it was like, what almost 1 year that between how to figure it out until you actually at second article came out and he started getting the details. Then yes, I would say yes. I think the Articles relation is almost one year. And in the meantime, it was becoming a buzzword in the desert community that there is something called a dimension. It was getting some attention, not to the level that we have right now, I guess? But yes, sir.

Yep, still we will we will share the article. I mean, we will share the article, the comments here, but you find the original article of datamash goes back. I think it's May 2019 and on the blog posts you had the realization.

Did your colleagues in your group also have that realization and and and and and what exactly is that realization? I mean we were saying the same things over and over again expecting different results. That's insanity. Is it more, is it deeper than that? Or was it just really that?

No, it's a I'm blessed to work with some very smart people who are much smarter than I am and they also have the same realization. I had very good support from my ATM business counterparts to explore the idea. It was not like this is that idea? And let's not pursue it. Not even 6. Let's not even exploded. It didn't I didn't say. Kind of a challenge but like, any new paradigm shift and tattoo of this, Mega, a scale. Come to think about it. There was no action on execution, so far. So, if I join the company, I'm almost in the second half of 2020 and previously than Urban initiatives to understand it. But I think the difference that made was, let's take the steps to execute it now because without execution nothing will change, right? You have to, you have to take the steps, you have to take the

You have to be bold about those things and without trying, you will ever know whether that happen or not. If you think about the time now, we have many companies coming forward with real implementation.

Back, then they won't hardly one or two companies who were speaking about any kind of implementation experience and it was all theoretical and that was one of the other, I guess. The challenging part about datamosh too many of the people even right now.

A lot of folks felt like this this feels right. It's talking to the right problems. But how right is it actually going to work? Or is it just the rewrite Omar? What would you say is your definition? Is it the same as I do for a stickler of the definitions that she has called the reason? Let's call Apple and apples, and not an orange, and not a banana or not a watermelon. Because it's a simple as that of the idea of the initiative. And then complain about the idea that they did not work.

So meet me have to stick to few things as much as we can and I bought something. And when I do complain about. Of things to her, like some of the definition of Haiti hardcore, so you need a translation in between but Essence should be exactly the same if it is required. And then I do you stick with the definition of a paradigm shift at the social, technical concept or, or a video of doing it. I try to avoid using the word architecture because people jump technology solution to the word architectural the time. And I definitely condemn people who say it's a technology initiative. It's not technology play. Have a great role to play in. No doubt about it. Very important role, but it does not start with the key.

All. That's why I love it. It's if you recall, you know, the when you look at how the organization informed right, you have data people process technology this

This better time. Use all four of them. And that's the beauty of it. You you have to move the scale on each of those characteristics to make a difference. If you want to make a difference, just introducing a stool or just changing the organization that have limited or, you know, those kind of benefits that you already have seen the boss Enterprise data, lakes or Enterprise data warehouse initiatives. And by the way, I would look up value in their times but things have changed, the people have changed their skills have changed technologies. Have changed organizations are becoming more and more data and are they want to be changed? So old Concepts have to be changed as well and that's why I

Light Rail to Trenton long answer to the definition questions, and sorry about that. It is early in its formative and I think that honestly a lot of good thought has gone into what those four colors are. What's are the konstrukt is. I know that actually doing a great job of working with you and any other leaders to relate to continue to test the ideas, kick the tires on it. Right? And I think that's really important that, you know, when you and everybody learns about me or talks about the Innovation is about that message clear that the socio part of this is very important verses, you know, some of the previous big trends like whether it's big data or a I or, you know, even date a fabric, right? They tended to start a little bit more, from a technical sort of inspiration or concept like

That sort of thing. Can you talk a little bit about how people in process, you know, the social side of this really plays into didn't rush for you, especially at the Admiral.

Yes, and maybe the best way to address would be to look at the four principles because three of them touches the social and the process part, right? So starting with that domain orientation, people coming from software engineering development site of things. This is nothing new. You are designing applications, using terminology is using concept using processes, which are associated with a particular to me. Now. I've seen people getting stuck on the definition on this journey. I might have done the same but thanks to some of our colleagues who have done some amazing work in this area. A shout-out to Daniel b. B overcome the challenge in setting lot of groundwork around to finding the domain send, keeping the definition simple and also making it different from your organizer.

Business structure chart necessarily, but I would encourage people even if they go with that because some organizations are defined like that. It's okay. Just chilling. Just go ahead and it don't stuck on the 1st, and it's not going to kill everything. And that's the first social part. There is no technology involved. You need to identify the right mindset business and i t l if it's an Enterprise shop you need. If you are starting with this journey, you are looking for people who are already, the Evangelist, the leaders leaders who will be willing to take on such a big thing. You need people who are able to be sources and find those right mindset, right minded people who will join the team that shapes the second principle of data is a product and having the right.

Protect demon face. And one thing that describes that is that ownership you need people who are saying,

I will own the data in my domain. I will take full responsibility. I will make sure that I ventured from operational systems. I make it available for analytical use. I guarantee the quality, and this is the, this is my quality metrics are associated with it and who's doing that that teams doing that. It's no longer the centralized ITT more centralized business team, or some other outsourced. When there, it's that product team and ownership is the heart of this whole day that my sponsor. That's why I called it in my previous talks that while no data of this initiative. And while you are structuring it around to me that's like a soul of the whole thing, but

Jordan, if you don't have the data product team, right? Which is all social, and then the skills and capabilities will come in. It's not going to fly. Right? And that makes a difference and even on the second principle and more social things like many other organizations, people are very familiar, with project-based set up. Doing things are starting throwing it through the operation site. Thanks doing enhancements updates to it and that again will not cut or find a productivity.

You need long-lasting teams, the right size teams and out of my experience. I can tell you there is no fixed formula of what right side. Steam is depends on your juice, is in your gratuity. And the value that team is creating, of course. And then mine said, it's very hard to find and Tim. I don't think I have to tell you. You leave the team a productive. I guess, you know what I'm talking about and that product mindset makes a difference that ownership of treating us, like your child, your baby. You're so cute rated. You nurture that you used to make sure that people use it and you're not protecting it. You are not protecting it at all. That's not happening.

Different, various different psychology you are going with. I want to create something that will be used again and again and again, and that's also drive a lot of value so far. Everything is social nothing to do laundry food and these other two principles, but this this is awesome. I mean, we've been already kind of 10 minutes until versation. I'm not even touch technology and I love it. I love that. We haven't talked to bring up and throw it back to you. If you said find the right, I T L look for the leaders were taking the risks, will mobilize resources and I heard this joke. Super funny guy when he has his jokes as you want to go. You think back into 16 people were staying up and down. Want to go go to the Moon. He's like no,

People wanted to go to the moon become astronauts. They're freaking crazy. So you need to go find those people who want to be astronauts who want to go to the moon and everybody think you're crazy going to go to move and do Sofia. Booster pumps are going to really change the world right there. So I think you need to be open to change right now. We're driving ourselves and staying at this moment about it. I got to find a domain right at all. Kind of car has to be something like I just start with something. What is your recommendation to actually get started? Like if if you realize our people are like getting stuck on 70 like I do.

Very tenderly. Like what should you put your somebody do immediately?

So the two things are connected. Once you find your crazy person who's willing to take that Leap of Faith with you. You will see that that person weather from business. All right, to your joint, right? It's belonged to some function or that one particular domain and that's your starting point. As you don't have to find 20 more to maintain. Seems you don't have to define the organization with all the leaders to find. That's not going to happen. You don't need it. It may not even be starting a new initiative, no idea what's going to happen next. Hardly any of limitation available. What is the entire organization is behind it? And you don't know what to do. That's the crazy part of it is rather take the first steps to learn from it and send.

This model specifically, you can go nuts on scalability and the fourth principle plays, a key role in going nuts, in are more organized fashion. You don't have to go nuts and destroy everything and then comes and then come back, the technology companies because you will need technology to support lot of things. No doubt about that. But once you get into execution, which comes to three steps later, you will need technology to do that. For an Ender eye, remember, talking to you, in some of our previous conversations. Be learned by working better Talk work, so they shall talk to them.

They thought a lot of things and I insisted to work with your mark to know what she thinks about the concept of how many times you get an opportunity to work with the creators, the inventors of the people who have come up with the idea. How cool is that like you are sharing that amazing mindset thinking and and you are with them on this journey is an amazing feeling to have and explore whether it's going to fit your organization.

Maybe since you don't mind this one, if your organization culture has no, I want to do everything. Centrally it would be nothing happened, decentralized? When I may not be the best fit for data mesh, it works because it's decentralized. You have certain aspects, centralized only intensify definition and there. If you have decided that yes, it's a fit. Then you start with it. Then we did what we call this accelerate Workshop, which is all about bringing that the crazy Bunch together.

That's crazy. But just now coming to same understanding about data mesh because I just think about it. If we have just read the article, we have different views on how on what have we that, right. So I remember people asking and challenging all in the data match for you. Don't need any any centralized the teams anymore. If you're going with domains be able to get Silo's because it's the way I know the answer to that is no connectivity. There's no shiting, all of those things. And how do you play on the same page with the people who have created the concept, you know, and I remember inviting jibaku speak to us. There was a huge that they are almost two hundred people.

Who came to listen to what she had to talk about? Then? He planned 90 minutes. It's meant for two hours. At least. That lady that makes the difference, right? Once you start a line in your what you want to deliver. What does this outcomes, what value you are pursuing, then the team don't act crazy, Bunch will eventually can decide. Now. We know what we have to do. Let's figure out how we're going to do it as a next step. And where is the data products will fit into enable those outcomes and who is going to be my user of those data product tubes. You figure that out in a step-by-step manner, as well. As any product team does, Jim correct, me. I'm, I'm not in your shoes, but don't you do Discovery, don't you talk to your cuz you guys don't you discover those things?

I'm so happy that that product mindset is finally becoming something that were talking about. I'm thinking about in data because I feel like, you know, so many bad patterns are formed around the day, whether it's like hoarding the data, whether it's in the data or everything is just a request from a standpoint things, like surface area, right? How do you manage the surface area of what you're doing here? And, which products are being used in which are not in the things that are not, maybe they should go away, right? Keep the things that are working and I'm so glad that we

So that is a good thing. And, you know, just a quick note from our sponsor. This episode is is brought to you by Davidoff world, for the modern-day to stack. Did It Out World next day to Discovery, governance and Analysis, easy turning data workers and tonnelle in superheroes, or you. I think this really kind of leaves. Do you talked about these different colors? You talked about the value of sort of date as a product. That's one that I think can be confusing to folks, because they're not as familiar with cameras in India.

So I was about to correct, you you missed a computational but you are very good at this because I'm in so many times and I asked this question so many times and no people can read it in the book that is out. And also watch it on the companion episodes on YouTube, how to do that. If you would like me who read books as well as left to waste time on YouTube or learn from it. I would love to do that. But the way I understood is that it's no longer one team, whose government things anymore from the process point of view, which means that let's take the example of human rights and navy, come naturally to me that I've been living here in Switzerland. Now, for almost 10 years. It's a federation. It's not like

Body controls everything and that's the concept over there. And that Federation governance is deciding on certain aspects for example, agreeing on that. These are the minimum things that are data product should deliver we have time for one. I think you would, at least your ABCDE checklist, which is also very nice. We have similar checklist as well. And in the book, you will see six or seven characteristics of data is a product. You can map them all together. They have nice overlaps. Nothing conflicting between all the three things and that could be the agreement or of that Federation government's from the process point of view. How are you going to enforce that? Are you going to leave everybody to?

And Willie published things on the famous catalog. And, you know, we are on a catwalk show. So we have to talk about that a lot. And I imagine that, that's the only way to do it. You don't have any Automation in place. And you have a nice day. Product, which has some nice data sets and it for the sake of Simplicity, put anybody you look at luggage, you have to manually meant Erick not going to happen. People will do it once. If it'll be, dead, disconnected, Destiny, catalog, and nobody using it, talk about the interoperability or the accessibility part or traceability that you can place how that data product is created because we still are bringing these out of our operational systems.

How those metrics are calculated which are exposed to my data product for certain outcomes that have user needs? If I need to expose that, I want my data analyst data scientist to find that metric daughter, kpin notes, definition and know how it is, calculated know where the different theater companies. Came from. Imagine doing that without automation. That's where the technology starts to come in that you attend, start enforcing those agreed governance principles, ranging from exposing how you will set those objectives that this this is what I will maintain a. My data product will be refreshed every day at 9 a.m.

On every one hour, it will have extra level of granularity. For example, each will have sales value in 5 gal seats or one currency or no guarantee. It'll have XY the dimensions and I'm just throwing things around. These are the things that each team is defining contable. Tell their consumers, that this is what I'm getting at. This is what I'm offering at the product and then the computation of part comes in. How will you measure? It is somebody will wake up at 9 a.m. For every hour? Check the freshness, check the completeness. That's not going to happen. You need those computacional aspects associated with it. And that's where the computer comes in. And that's why it's not just that traditional governance that the board that I don't like at all. Why?

I am looking for a better world. If you guys have any idea or any listener or viewer has any idea of last 2 years that I started calling government bodies forums. For example, I don't know whether that will stick or not. But you see if he need to agree, what decisions will be decided within two men, and 4in interdomain dependent. She's busy with going agree on certain things. So a nice balance of full autonomy and where to go when you have to solve things together. That's all combined. Are some of the aspects that I would come to this part of that Federated computational governments agree. I think I've seen plenty of data mesh, you know, videos and folks in the talking about the Federated governance and then, and they and they skipped the computational part, right? And it's important to think about how do we

Automate in crito, me, right? And governance, as code and policy is code. Is that something that you think is an important part of this too? Or is that just one example of an implementation, some of the things I've implemented as cord, where it makes sense? So, for example, of that catalog again?

Why leave it as an optional step for the data product teams? Why can't that third principle, guys, and train technology topic about to happen, high alert. So that will enable charging capabilities. One of them could be as part of your data logs pipeline. Maybe the last step, is that automation piece that you enabled by a court. You help every domain 2000s, this taken care of. Don't worry, about automating things into your catalogues all your data lineage, worry about the things that only matters about your product. Like, describing it not acknowledge, even describe it. Now, you have what you have put in that field. Call Amor table. At least not right now. It might come up with some ideas and it looks like it's a email, okay.

Customers and employees context. Oh my God. I'm using fancy. Words is something which has to be defined and that's that can be embedded in a court as a final step. So you are not leaving things are optional when somebody will push your pipeline into production, you know, things have been enforced, your platform might be able to identify your columns and identify pii data things which needs to be encrypted or anonymized excetra. Excetra. We have some examples where we're doing it. So, the technology is enabling things and make life easier, even in a very decent realized work.

That you don't have to reinvent the wheel in each to win and each team. You can still followed lot of agreed processes and still have full freedom to do today.

I want to get a couple more things because I mean time, look at this. Oh, I'm so excited. One is thinking is kind of a very, very crucial aspect to your end. And you mentioned like, there's things like Fair like him and I've been working on this whole ABCD. I love to get your impact that your input on this like a boundary is a contract and expectations for your Downstream. Consumers, explicit knowledge. Fair is final accessible in her up a bulletin and the reusable Central Point of View and that should be the central, you say, your decentralised nodes in the message. You got this, you got to follow this thing.

What, what is Omar's ABCs or fairs or or what would you expect and what it, what are we missing from? The stuff like that? You and I have been talking about. What else? What you get for cycling War start.

Show. Going into very specific things because nothing comes I can't recall. Your ABCDs. Fear is something which is reminded to be a hundred times a day, but in general, but I would love to encourage people is that not just tell people. Hey, you need to make sure they don't fear. What does it? What does it mean to beg their affair? Right? What does it mean to make a definable or you have to publish it in a Catholic.

Do I do it manually? No. No, you don't have to do. It manually. Here is the automation Here's the final steps. Once you went to your data product, team is given access to your data or environment. You will be able to do that. I think that's how bad is very important. How bad is that platform? Beam? Who's also managing their platform is up to its beard product feature to enable its for their consumers and users to use. So, I love that product thinking applied on the platform as well, where they are ensuring that the data product teams and the different team members and their different personas like, engineers and developers and US developers. And designers, a product owners product managers architecture. Most of them will be some capabilities of the platform.

Swim teams also using that product mindset and it's not just about you have to make your data findable that that platform team is actually teaching on during the part of their own boarding. Slashleen Inception. Let's call it that used to protect them during the lean Inception. The team is guided how to do it in the beginning of our journey. We had minimal Automation in place. I'm proud to say that because now there is a lot of automation that we have done it, just helping us to scale for their in multi have a multiplier effect 360 right now. And we are trying to go as much as possible with even more automation, where it makes sense. Not everything, didn't some control is then left for further Innovation and Improvement.

Is this is a great point about the, the three more of the social ones? I think the fourth technology. And actually, you know what? I don't want to talk about technology. There's so much technology stuff. This is not going to call you one more thing. I want to hit before we go to our lightning round in past conversations. You were telling me about this, when you called your acceleration Workshop that you had like a it was a four-week process. And if you got the people together, they start at a high-level. They got it. Did a deep right in it. Please explain what this would because this was this eye-opening for me, please listen.

Yeah, so I'm not an expert. I've learned there. So I'm going to just chill by experience. I did then came back on. Came up with this idea of the acceleration is to bring all the stakeholders together and you narrow down from the whole world of white Campus of not doing anything. But we are going to do to Bringing people on the same page, focusing for that particular domain. What are my object? If the business objectives that we have to deliver and how am I going to measures? But that I have delivered those

Objectives are. In other words, you are working with them in the context of your data Machine Shed, is defined in your okrs night, Belgian to that process, just like things. How will you achieve your okay. Are you are cute. As long as you have some type of this is some ideas that maybe idea 1234 did, this will work and you don't know that, you don't know it yet and then you will identify what kind of things you need to do and deliver those ideas do stuff that you will do, basically going one level below, and then together with team.

Which are we going to focus on one or two particular use cases? And if you have the right people in the room, those crazy Bunch, which has the right domain knowledge. Who knows what's keeping them up at night or trip business problems. They're trying to solve.

Didn't know what the problems are. They have amazing knowledge in this area. They are looking for guidance. In many cases empowerment. Let's do it together and they know how to prioritize in what to prioritize because even on the fourth day of that Workshop, people start to link things together that if he picked used his one and two, we will enable even you skate 3 and 4 down the road because we will be sort of the foundation work off those data products that can be used for more use. It is not just within the domain, but outside the diminishment.

Right. That's that's the starting point. After that. You do your Discovery, you pick up on yusuke's, you identify in the context of the data who was going to used to do those tasks. A sexy. You're talkin about end-users, you're talkin about those. Personas your talk talking about what kind of inside they would need based on certain assumptions to deliver those doors inside those activities, which they might see it as a dashboard. Light years are the board. Might she might feel something by email or it's embedded in some important, maybe somewhere.

That's what Jen. You be outcomes that are expected and you can start walking back to words. What kind of data set data product will be required to deliver? Drop G. You don't know what feels like, you don't know what tables are there is that you don't have to go into that detail. That's the first and that's not happening in four days. You are going through a long process, especially in the big things and everything is new. You make my dicks or weeks. But once you get mature, you are doing this as an ongoing basis, you may agree on some kind of a discovery Cadence, which might be triggered by a feed bag of your users. Might be triggered by some other events, might be triggered by change of people. All of those things will figure that once, you know the systems. Once you know where the day. Might come from, you can definitely identify and Link the pinpoints of either getting that data the gaps in the systems of gas in the

Metrics and the KP eyes or even complete Gap that I don't have access to any of that data right now because we don't have an underline system, but we don't have we haven't purchased that data that site that goes on, in terms of acquiring things shaping like the whole cycle of curating, things generating inside.

Giving it quickly in the hands of those users. He did this. Help did this book? Is this what you're looking for? It's like taking of vertical, slice of your four layer cake for, exactly. And now I'm ready Company by Justin King about that case, but that vertical slice in the world in 4 weeks. And then the next day, then the next day, it's too late. That's not going to fly in my view. You can try. It. It doesn't work there. Usually take some time to develop deliver. That is some kind of the cycle that we have now that we are trying to practice in various domains and have varied level of success and learning so fast.

Yeah, they did it this week. Will you call the vertical slice when I called the iron thread that you want to go do like from start to finish figure that out and at some point like you want to you decide, I'm going to make that thread stronger. So, you know, it's true because you probably had another source whatever it's somewhere else. You'll going to go get another thread and then those threads start to all right, it's time to go to our lightning round. Lightning round presented by day. World. The Enterprise data catalog for the modern-day to stack. So I'm going to go. First question 1, do you feel that? It's important to approach all four pillars of date of national tennis League.

Or is that playing? You know, it's yes, but you don't have to wait for a perfection in any of them.

You use a controller like a, like a scale, right? The oldies can see it in their cars, you know, you had a volume button up and down or what is a or b? C artists, like a volume on All 4. One will unfortunately have limited success. And product parlance. All of them are acceptance criteria. All of them are requirements, but you can do minimum viable of each of them.

Yeah, so you may just come up with his. Let's just agree. We will need a name of a data product and the description wasn't doing Allgoods. Let's define the next door next month.

B b, y u by baseball.

I love it. Straight to the point. I love it. So I can question you talked about their principles. You talked about. You mentioned, our data, ABC's, the date of product ABC's, do you think that these principles are something that management should mandate? That should be company policy.

So, sticking with the data mesh principles Federated computational governance, that Federation of the people that you are working with in the big thing that crazy Bunch. Let them decide.

B, y impose anything on them. They are. So they are selling their product if they have agreed to take ownership. I'm pretty sure they will stick to load of principles, not just these four, five things. If they have, the right-minded architecture there, they might even agree. Like, I insisted on having a proper data modeling architecture. Like the Doubletree that we are using for the data set, which are you going to apply update about tomorrow? Not to buy the ocean, right? So why not let them decide.

It was good one like that. Question. Do you believe company should hire data product managers or just adopt a product mindset?

So it depends, right? Do you have a crazy bunch to have that crazy person that person who is dying to show? Yes, I will make it. Work is the team willing to learn. You can invest on some nice productive meeting. I love the word from Silicon Valley, padur, grow, big fan of what Marty Cagan and he may have done with empowered and inspired. And now with the book love, for example, you want to learn Discovery go and learn from Theresa amazing stuff, right? You want to learn about? Okay, I'll stay so much available. You can learn from reading a book, medical photos of things like that. All of this content is available also in two different bags and shapes and formats. There is no limit to it. But that mindset is Keye weather. You bring an external person to do it at, but that person needs to be taking that ownership or best.

We didn't even further. Good example, some positive vibes of what is possible. I love that answer. People hire people. If people have the potential them, let them wear the hat right for you. If they ate a mess was never invented in Asia. Mike never did her personal blog posts are, would you implemented something like that or was knowing it naming it and recognizing important. I don't think that's what you're such a structure thinking, but it would have been some of a mixed form of what we have learned from implementing data Lakes, what we have learned from implementing data warehouses. And what's the next Evolution? Put look like they might have come up with a mixed approach.

You remind him come up with some kind of a lake house is it's called sometimes you might have come up with something completely different. It's quite possible that one of those combinations would have been taken but I would have definitely and insisted on avoiding. That's not been yet. Another Enterprise Lake. Let's not build another Enterprise data warehouse. We have had limited success. So if you want to really go take the next step, let things different see that stink again.

It's at some point. We are acknowledging that we are insane. Keep doing the same thing over and over again and expecting different results. That is one of my favorite things. And also we don't know if we're going off to our mesh minute, which is just going to ask a lot of people. You got one minute to rant pontificate about didn't mesh. We have done it already for 53 minutes, but and here's your one minute like you go for anything. Freestyle freestyle.

All right, so I highly encourage people to explore it. If it fits your organization Archer, find that crazy Bunch. Take that. Yes, you can do it, don't do to stick with the norms. And then in the comfort zone rights, go out and unleash the potential of the data. That's all it's all about. It's all about the usability. Don't take mine sets for granted. It's not, not everybody is willing to do that. So find the right people, if the people in this is there, which is important help them, trained em, and reap the benefits. That it's all about business outcomes delivered, so go for it.

Thank you so much. I was that was you have today, given everybody who's listening, a road map, based on those pillars of the very structure of your framework here to go through it. Very practical. One, and very important. We did not talk about technology and nnnnnnn we should do. We should be, I be interested. Can you do the song snowflake? Yes, you can. Can you do it with the virtualization? Yes, you can. But then we get into, like the honestly, I like I'm not interested in. This is the game-changing part deciding which technology go use. I don't know that person away from this is a change, right? This is what's getting me out of insanity. That's that's that's a topic for another $1. It's it will play an important role. I don't want to

DM precise this but without what we discussed today, there is no point talking about technology. No point none. It's not what I asked you if we could probably do an impromptu Carolina session with you or will talk about technology and then Omar amazing conversation today. I think some key takeaways that I got and then I'll pass it to that. You know, you came upon the idea datamash at first a little skeptical, but as you learned more, you really came to really like and appreciate that framework. And then they had the opportunity to really clever. It was your mom. And those articles came out in 2019 and in 2020 was the second article and the first journey was really a step of the journey was really realizing that, you know, doing the same thing over and over is that

And expecting different results is the definition of insanity, right? And made a mess. And in general, looking for a new way, forward was the right step being open to exploring that idea and recognizing that there were problems and this can be a solution to those problems. So I think recognizing the problem is really key here. I think a lot of folks are trying to get their organization to recognize the problem. It at the end of a jewel, is for that. And when you talked about your definition for datamash you, very importantly mentioned that it is a paradigm shift. It is social and Technical. It is all four pillars of the right and that it, you shouldn't talk about it as architecture. I couldn't agree more with that, because I think that when you say architecture or you say things like that, that really mean more towards the people jump to technology. They assume I need to invest in Ohio by a catalog.

Call that, right? And that is not the truth. But I think towards the end of your lightning around there or you're getting mad at you, three things need to work together technology to one process working together. And and that is that is so so important. And then you talked about domains and the importance of it's not necessarily about your org chart, right? That you can start from there. That's fine. I do it. Make sense for you. Find the right leaders in your organization. You want to take risks in those different domains and treated as a product, right, which is really important to me as as head of product that you said that sort of product ownership is the heart of datamash structuring around domains and building those two main products of the soul and that there's no fixed formula, but if you think about life,

Hey, what's the right thing for me? In terms of the size of the centralization and centralization? You can figure out the right thing, so much good there. So I'm looking at the Federated computational and she said, it's a way to go nuts. In an organized Manner and agree on what the minimal things are about to get a product and you didn't your light on your seat. Like I mean just let those crazy bunch of folks. Come up with it right there by the house. We go do that, right? What is findable put in the catalog. Okay, how do I do that? Manually automatically are the tools like this is what this is, what this is more than formation need to have the computational part, right? I mean, read, understand. How is that going to be in force? Right? You can't expect that things are going to go work in a calagator. Don't have to be manual. Understand.

A lineage this automation is where the technology comes into place at definitely plays a very crucial role in here because we're going to be the basis of our government's. What is going to be in a centralized culture? Freedom issues. Probably not going to work for you. And then this whole last part that we talked about the acceleration Workshop, like this is, this is a goldmine of very specifically what you go do, right? We talked about it. You can do it in for four sessions 4 weeks, or whatever, but get your stakeholders together, get the people on the same page on that domain. What are their business objectives? We can get measure for, okay, ours for this stuff, right? And then just like you do Kiara.

Navigate to one level deeper, go deeper and deeper right at the Depot. Realized those ideas. Either the most practical ones for the use cases. By the way, the things they are keeping me up at night and the windows don't mean to actually know how what, and how to prioritize. And realized we should focus on youth case to and 4 because that's going to able than 578 and 10, right? Then I realized I was going to go do what right? Talk to those users, which is going to help us to figure out the requirements and figure out what the sources of these days. We don't need to go into the details of what are the tables and columns yet at this big vertical, slice of what I was calling back that I earned thread. And again, this is a cycle that goes in through right and find the pain points, right? There's going to be gaps in the entire process, right? By those things.

That was a lot. How did we do?

Awesome. Be talked about buying to be. We did not talk about the body via the heart and soul will live. What we did that show, the photo of the Fantastic loved it. Let me throw it back to you. Quickly three questions. Where do the same? What's your advice about data and life? Whatever you want? Brought on purpose s. Who should we invite an ex or what area, whatever in 30? What are the resources that you record? What are you reading? What are the books where the conference is? You're going to witness one of the resources that if you're going to share that with everybody who's listening.

Okay, advice on life, try to be simple. It's not easy, but it's simple to be happy and it's very difficult to be simple. I looked at that principle about invite invitation. I would love to hear fellow practitioner. I'm not I or my team cannot be the only ones, so I wanted I would encourage people to come speak. Share their experience. It's a community effort. Learning Community is one of the best. I've seen out. There are Ten teams doing an amazing job. I love the passion. You, what am I eating? I'm just about to finish the remaining chapters of inspired which have been pending for a long time. Thanks to you borrowed. My knowledge. Graph book is here. I thought too late that afterwards that's on the list and I love her that the government has made me read the books once again, but I'm a Big Freedia.

Lot of dogs on medium. For example, I read a lot of Articles being published on things like towards data science. Call big fan of learning online with LinkedIn. Learning coursera's, data can stop this world these days. It's amazing. So it's up to you what you have to do.

Thank you so much for this and we'll, we'll, we'll get all the links and stuff that you're doing. So we can share with everybody Omar. Thank you so much. This is a phenomenal conversation. We were in so much because you're so practical and you're honest and no BS about this before. I truly appreciate it. Looking forward to meeting. You in person a couple weeks at Snowflake and it is very quickly. Next week. We actually have Brett hurt, who is the CEO David at World of Warships. You release the new book about entrepreneurship and talk about the cultures of kind of what it means to be a record before benefit Corporation in the date of space. And that we want to always think that it up world, the Enterprise data catalog for the modern-day to stack for supporting us. And have a long talk this week, at the news, every Wednesday have amazing guess, amusing conversation and

Omar. Thanks to you. Cheers and talk. Tell a special thanks to Dana. World for supporting the show, Carly berghoff for producing John Williams and Brian Jacob for the show music and thank you for the entire catalog and Cocktails and face.


Some of the greatest breakthroughs throughout human history can be attributed to the biotech field, from life-saving vaccines to advances in supply chains. So it’s no surprise that biotech is also a fast-mover when it comes to innovative socio-technical paradigms like data mesh.

In this episode, Tim, Juan, and Omar Khawaja, Global Head of BI at Roche, break down the good, the bad, and the ugly of the company's data mesh journey.

This episode will also feature:

  • How to identify potential use cases and pick the one to prioritize
  • How to think about ownership when data and tools are widely distributed
  • You’ll be at Snowflake Summit in Las Vegas next month. So will we. What’s the best buffet food?