View Recording: Fabric’s Place in the Modern Data Estate

Nathan Lasnoski 0:27 OK. And we are live and we are going to be having a great conversation today about fabric and its place in the modern data estate. And I’m just super thrilled to be here to talk about it. And we also have some other special guests. So this is gonna be a great conversation. We’re going to talk about fabric. We’re going to talk about how it aligns to an estate that you are probably building right now or one that you have part of and are looking to evolve. This is going to be a particularly frank conversation, so I hope you’re ready for it. I’m going to tell you how it is, or at least what I think it is, and I think you’re gonna have a lot of fun. So be very open to put your questions in the chat. Let’s hit it as we go and I’ll just start by introducing myself. So my name’s Nathan Lasnoski, concurrency is chief technology officer. I’m also a Microsoft MVP. I have just been really excited about the progress of fabric over the last several years. It’s something that or several years single year and it is something that has moved from a twinkle in the eye of Microsoft to being something that is truly changing the market in the space of data. So really excited to dig into it today and then in addition to myself being on this this call, we also have Brian Hayden and Brian Hayden is a solutions architect concurrency you want to say hi, Brian. Nathan Lasnoski 1:58 Awesome. Good to see you. Brian is gonna be around to answer questions in the chat and can poke in where he has some comments as well. Alright, so let’s get started. I I think this is probably something I’m going to want to sit on for a minute to let you kind of consume it. I think it’s important as we start talking about Microsoft Fabric, to do a little bit of a history on data states in general, and I think many of us have lived through these different stages. And when I first I, I’ll admit, when I was first told about Fabric, I was at the Microsoft MVP summit about a year and a half ago, and I was talking to the person on the lead of the team of fabric before it was released. And he talked about this cloud data state and I was like, why are you doing this? How does this relate to a data mesh? Why is this important? And at the time, I didn’t quite understand why this was going to be a good move for them, and now I get it. I’m hoping to convey like why it’s an evolution and why it was brilliant. And then we’ll talk through the different components. But let’s start with like the history and how it relates to other things. So where I still find a lot of companies, believe it or not, is in this this early stage, which we call just exporting to excel. It’s just like spot where you really don’t have a mature data warehouse of any kind. You don’t have a position of uniting what data means your organization and providing clarity back out to the organization in the business and you have various teams working within business systems. They export that data from the business system to some form of analysis tool like Excel and then they take action on that data. But nobody trusts the data after that has been exported because it’s lost in time. It’s like just it’s as good as when it was exported and the shared understanding that’s necessary against that data is very limited, but it’s something a lot of companies do and it holds them back. So something we’ve always perceived to understand is better than that is to create some form of data warehouse and many of us live through these patterns of building a data warehouse on premise with ETL practices and SSRS reports and collections of report writers that built different reports that were then issued out to sets of consumers. And that is some cases was successful at least creating some level of agreement. But in many cases it wasn’t too many cases. It spent many years building a data warehouse that UH was limited by its ability to bring true value or to take too much time building data warehouse and not enough time actually getting value to the customers. So it oftentimes was struggled with that, but sometimes it made it all the way and it’s providing value right now. And I know many companies that that’s the case for, but we evolved past that and we started to move into using cloud patterns and cloud patterns move to ELT versus ETL. They usually follow to bronze, silver, gold pattern. They’re popular. Examples would be Snowflake or Azure DWH. You Azure DWH Gen 2, umm, usually surface through power BI and umm, some collection of potentially tableau one point in its history and particularly the first snowflake. This is when it just took off like a rocket’s right. Like people are like, I am gonna take my data warehouse and I’m gonna dump that thing into a snowflake cloud environment and then you can go to the ball. And that strategy sometimes paid off for people, but more often than not, they moved a lot of stuff. And they got a value from the small fraction of it and a lot of the stuff that they moved just kind of sits there. So there’s a question now with many companies of did I get value from that? What am I paying for? My snowflake implementation, is it really doing the right things? Did I bring the data together and gain alignment or do I just have data sitting in another lake that was from a bunch of separate ponds and it has some of those same concerns? Now it moved past the idea of generating this like slow moving behemoth that was the old ETL data warehouse. But it also generated some new problems around getting value to people which is when other concepts started to take hold, such as what data bricks has been very successful with on Azure data Mesh Concepts Lake House concepts that have started to emerge, the data mesh pattern in a very practical way allowing us to use an architecture but focus sets of concerns and bringing that through the cloud ecosystem to a set of consumers usually being consumed through power BI and another place where there’s just tremendous momentum and forward movement. The challenge with it still remains though, and is something that’s kind of existed at all these stages is you had to care about the infrastructure and you had to care about it a lot. Like, there’s still a lot of question of even if I’m building on a data mesh pattern, WOW, does that mean I have like 7 data warehouses now? And how does that all relate to each other? And if I’m building on data breaks like, am I not learning data bricks in addition to Azure and addition to this it’s just creates a lot of that additional question marks. So where did? Where did this kind of take a next step with Fabric. Fabric starts to bring that data state forward toward a position where we’re moving away from some of the slow ETL processes. Even the ELT processes to assess platform where we start to move away from having to think about the infrastructure components of how it served. Still delivering mesh concepts and doing it through focus on real time mirroring of operational databases into the fabric environment that then can be used potentially still in bronze, silver, gold patterns, but getting away from the sense of like I’m running a job that job takes X hours and then it finally gets to my location where I’m gonna start doing something with it and not that some of those stages don’t exist after that point has been achieved. So it’s very interesting how the cloud is starting to simplify the process of us using data. Maybe get us to a point where we can bring that data to our consumers faster with less deltas between when we generate it in our operational database and when we can realize the value in our data state. That’s what either data scientist or analyst that’s taking advantage of it. But I think that’s not really the only piece of why this is important. So I thought like connecting that history it would be worth giving you a frank discussion on why this move was brilliant from Microsoft Point of view. And again, this took me a little while to arrive at in. It’d be interesting to get your perspective if you still think it’s really coming out the door. So here’s why I think it’s brilliant. Microsoft has totally dominated the consumption layer, so if you think about the consumption layer being power BI. Tableau had its time just like a lot of other things right now. It’s a dead platform walking. If you’re still on Tableau, I’m sorry. It’s over. One Salesforce bought it. You kind of went into a position where Tableau started to really lose its relevancy in the greater market of skills. So Microsoft has done a great job of dominating the layer that people are using to get value from data. Now, that doesn’t mean that like the consumption layer isn’t also data science, which we’ll get to later, but from analytical consumption layer where you have just huge numbers of business users getting value, building their own reports, creating their own at capabilities, any consumption layer to analyze and see that data. Microsoft has done a very successful job of dominating that market, especially the legacy sort of analyst space. Just to kind of reinforce that point. Tabulous kind of in a position where it’s not winning the the the the race at the moment. Just move off O in the cloud infra game Microsoft was having to compete head for head with other partners or other companies. So you think about like, alright, Microsoft’s got the consumption layer largely dominated and then it gets into this competition at the infrastructure layer of or your data state going on Azure is your data state going on? Google is your data state going on? AWS maybe all of them and what do I do about it? Is it Azure? But actually Snowflake on Azure is it Azure but actually data bricks on Azure. All of those conversations are happening even though they own the consumption layer, so they own this consumption layer and they’re like, man, like, we’re still fighting with where the data goes, even though we owe we own most of the consumption layer that people are using to access and work with the data. So by making a SAS platform, what has happened, even if it’s not as feature rich and I’ll fully Amit that at the moment, even if in some cases it’s not as feature rich as some of the other platforms that might exist today in the data like movement space, their mediately made every power BI user a fabric user and that was brilliant that move of enabling the commodity side of data engineering for the power BI users was a great directional step. And the reason why is because it enables more people to be able to get into data engineering that may not have been individuals or a comfortable building in Azure estate with a subscription with the resource group, with the deployment of synapse, with the deployment of these Azure data factories that are moving data from point A to point B and all that stuff they’ve taken that and said, yeah, I know you still have to do with data movement. I know you still have to do like cleansing and things and labeling and everything comes along with it, but. You can do that in SAS platform without having to worry so much about the infrastructure and by doing so they shifted the whole conversation into this SAS world and that’s creating a strategically simpler use case and approach than what you’re seeing from a competition. Instead of standing up an Azure data warehouse and all those components that we talked about or a snowflake instance or putting in GCP and then trying to get it so you can view it in Power BI, which for a technical person, you’re like, well, whatever I can do that. But for like 99% of your business users, they have no idea how to do any of that thing. Those things, giving them the ability to surface data from business systems more quickly is really spark. And then if you see that integrates things like copilot into fabric for data analytics and starting to change it from being a conversation just about surfacing a dashboard to surfacing data to those business consumers. But data in a way that they can interject and talk to you and have questions of and interact with and in variety of different ways was a really smart, really smart move. So I also appreciate how they took this story and have them integrated Fabric with a whole variety of spots that already exist in the cloud. So the fact that our business systems are moving to the cloud. Allows Fabric to build real time connections that don’t have a negative impact on the operational platform, but mirror the data into fabric and allow it to be used by the consumers to be able to get value faster than what we would have had to do in moving data from the on premise environment through an ADF pipeline to a data warehouse to a different stages and then finally to the consumer. So they’ve made that. I think this is where like Fabric is truly going to win out is not just that they own the consumption path, they own the drop down into then doing data engineering in that same world, but also our facilitating data transfer capabilities that would have been really difficult to accomplish with any of those legacy approaches with possible exception of like what data bricks is doing with some of their work right now. Data bricks is like totally tied into fabric, by the way. Like, their path continues forward in a really interesting way. So now that I understand how that all fits together, I can see why a if you’re building a data state from the ground up and you have any desire to minimize the amount of work that your teams are doing, Fabric fits into that picture and a really substantial way because it allows for you to offload some of that, just like maybe talking about Office 365. I remember when they were first bringing this topic up. They were like Nathan, I’m building the OneDrive for data and it hit me like OneDrive or data. But totally locked in this idea of like, why that’s so much simpler than the way we even dealt with SharePoint before. Like going back in time or just like with the same way that like teams simplifies using content in SharePoint in a sense and then also tells you how much it’s going to create sprawl and other problems. But it’s definitely aligns that idea. So I’m I think it’s really important to understand Microsoft is taking the path forward to create truly assass ecosystem and then also creating an opportunity for more people to get involved in a powerful way in the world of data. So why not assess data state? I thought about this and sort of a sarcastic way, but I think it’s really in a sort of non sarcastic sense. I think it’s easy for us to picture why you cannot have to use the SAS data state because for some of us doing these things are easy. Like I said, building a cloud data state is totally easy for some of this. It is for everyone’s called data States, understandable and governed, certainly for some of us it is. We are great at optimizing the cost of cloud data. I’ll walk into a company and see their data state and know they could be using reserves. They could be simplifying those costs. The security layer is totally taken care of. Many companies it is, but I think what you can start to see here is that for most companies it’s not. For most companies, it’s not. So all these things that like we could say with a straight face, they aren’t easy, they are implemented, they are optimized, they are governed. But if you take 99% of your companies, they’re not. It’s not easy for them. It’s not easy to make it understandable and governed. It’s not easy to optimize the cost. It’s not easy to optimize the security layer. All the data still is in the ASK for a hunter, so these these situations put us in a position where ASAS data state for many companies is just straight up the best move and it puts them in a position to be more supportable long term without having to have as much of that concern. So the reason why I think the SAS data state straight up makes a ton of sense is I think it eases the value of getting eases. The process of getting value from data, the simpler we can make it, the wider the tent can be that we still govern. Something like Fabric can truly provide a value. A effective path for our data to get to the hands of our consumers without needing to care as much about the underlying infrastructure that sets it on it. It also allows us, and I think this is something I’m already seeing and you should just be knowing yourselves. It provides a tremendous access to innovation that’s built with faster time to value. Microsoft Data team is totally focused on Fabric. People like they are putting all their energy into this and you can see that by all the landing of AI innovation, copilot platforms and such that they’re starting to light up within the fabric interface and is already a known tool to the power BI users. So it facilitates that that ease of access to innovation because we’ve done the prep work, there’s, there’s this idea of you like when you get a new light bulb. I like we’ve been having to switch out light bulbs, right? Like you get a new light bulb and you just screw it in the same socket, right? It I we went from these like old light bulbs to like these. Like there was the 4th, the phosphorescence in the middle that spot. Now we have these LED’s. Now you have LEDs that change colors and out enabled. They still all plug into the same generics twist in slot right. Power BI is kind of like that, right? Like everybody from the consumption level, everybody’s generally using power BI from the standpoint of notebooks and other things that can plug in too. So like the consumption level kind of stays the same, but the ease of using the underlying infrastructure has gone a lot simpler. So I think that’s facilitating that ease of access to time to value because you’ve brought in the set of people that can actually use it. I’m I think another piece of this that’s really critical is less confusion over data lineage. And I’ve got a couple of screenshots of that later, so I data lineage has been like something that’s always been an add on to most of these data States, I mean, yes, you can see it in the context of ADF or other places to a degree. But understanding it end to end always took you applying per view as like a secondary tool into the picture and it felt like this very side add-on and per views catalog was always a little weak. Umm, what I like about where they’re doing is they’re kind of tying purview and fabric together in a way that provides a much more cohesive experience in data. Lineage itself is a critical need just in general, so having less confusion around that path I think is really important. Umm, fourth thing here is this idea of real time access to oil TPP systems without slowing them down. Many of the companies were working with are companies that, like they have huge real time processing needs and they then need to extract it from that to be able to use it in the reporting purposes. And one of the things I’ve been most impressed with on the Fabric side has been this idea of mirroring those systems into a Fabric replica that can be used very quickly for the needs of the consumer. And that’s not something that was really possible with many of the on premise systems. Historically, umm, at least to the degree that you can do now and then. Finally, I think cost optimization is going to be a big win here, like building an on premise system to do some very simple things, but needing to have it burst up for certain use cases, then having to move into building up like synapse systems that require like very stock implementations of capacity as you will see you later, there’s some great opportunities to build on capacity across the board, like being able to gaze to get a set of reservation of capacity and processing and then be able to use that for a variety of purposes across that Estate. And then sort of refill it as you need or get more capacity at lower discounted costs is going to provide a ton of value to companies that like don’t wanna pay for the big infrastructure or don’t even want to pay to have like that middle tier infrastructure that they the alternatively need to to set up. So as we start going from here, I would love for you to continue dropping questions in the chat. It’ll just take a break to see if there are any. There we go. Cool. Thank you. Nice question, Jamie Gunn. I’m just double checking for enterprise. One challenge that I’ve seen has been more on the data movement aspect for your PCM destinations, where the data can be consumed. Line of business apps or customer facing websites, for example near real time replication versus batch, said external folks that tend not to trust the data would love to hear all Fabric help solve them. Solve some of these concerns. Thank you for that question. I will address that as we go through and Brian, feel free to come off if you have anything up to this point that you wanna come down, but don’t feel that you need to. Brian Haydin 23:45 Yeah, James question was excellent. And I think you kind of touched on some of those topics, but I know you’re gonna get into it a little bit deeper. So but keep the questions coming. Nathan Lasnoski 23:55 Cool, cool. Brian, if you want to respond to that also in chat, you can please do. Brian Haydin 23:59 Well. Nathan Lasnoski 24:03 OK, I thought this was uh, this was from a Microsoft deck, but I thought it was great, this idea that I’m the Chief data officer. I don’t want to be the chief integration officer. Every CDO I think this is representative of the hardest path in creating true data outcomes is getting a agreement around data and B tying the systems together to be able to achieve outcomes from data. Some of that I think is unavoidable, even if you’re in something like Fabric, because you simply have to gain agreement on what the first version of the truth is. But I think what Fabric is going to do is help us with lineage and help us with being able to pull data together in a way that is less dependent on multiple data platforms. I also want to note this is I think, a really important point. Surprisingly, Microsoft has built in Fabric, real time integration and not just into Microsoft. There’s a fair amount of work they’ve also done into Google, AWS and Snowflake. As part of this platform, that kind of shocked me in terms of like what they’ve done in the past. So I think you’ll find it very interesting to see that if you do have pawns of data in different places, they do have this idea of called shortcuts or mirroring that will allow you to take data from these different repositories and mirror them real time as well. Or a shortcut them if they already exist. OK, so wait, if Microsoft has a SAS platform, what happens with all the stuff we’ve already been building? That’s exactly this point. So the goal is for existing cloud platform resources to mirror them into fabric. So if you already have many of these platforms, and especially and I think this is kind of like a really important thing to note on many of the times, if you’ve spent much time in the idea of a data mesh like this has been like the topic detour and it’s really what Fabric is built on the idea of, it’s just really, if I were to boil it down and said data has separation of concerns like the idea of throwing everything into a lake has never really been the end game that we have, separations of concerns and. Enterprise and we build those separations of concerns toward outcomes that need to be realized within the context of data, whether it’s data science outcomes or it’s data warehousing outcomes. And the goal for fabric is to facilitate those different sources of data, whether it’s a SAS system like a CRM platform like Salesforce or CRM, Microsoft CRM or ERP like SAP or something along those lines or dynamics ethno that lives in the data verse or another data warehouse. All of those represent different parts of a data mesh that aren’t going away like you’re not. You may have a part of your application that’s running in AWS. You may have part of your application that’s running in Google or in Azure as a full stack application, but we need to get that data into a data warehousing platform that’s still respects the separation of concerns, but lets it bring that data together. One of the goals Microsoft has had, and we’ll talk about that with one lake, is even in the context of separation of concerns, we want to store as you copies of the data as reasonable, and we can then abstract that into the mesh separations of concerns without necessarily having to duplicate it, which is what a lot of mesh concepts have been historically built on. Is this like publisher subscriber model of creating copies of things which just ultimately kind of slows down the process of getting it from point A to point B. So as you explore more about its capabilities and the separation of concerns level, I think you’ll find that that really complements why SAS platform that sort of floats above a lot of these separate little instances of data is really helpful. Umm many of you are data bricks, users, data bricks is like on fire in a good way right now. It’s aiming to do the same kind of cool stuff in fabric, right? So it’s it’s a partner of Microsoft. They talk about all the time. There’s probably a series of whole conversations just about how you use data bricks in the conjunction with fabric. That’s something to go talk more about as well. OK, so governance, how does governance work in this brave new world of yours? I had that same question. I talked about that a second ago, per view Fabric. World sort of non congruent for a while. A very fascinating like when they released Fabric initially per view couldn’t even talk to it and I was like, what the hell, what is going on here? Like, how does purview fit into this picture? And now Fabric has full lineage in the context of its own interface and per view is linked into that. There’s some very good talks on Microsoft Ignite from last year, but more that we’ll be released especially at build around the relationship of purview and fabric. I think those will be great to understand per view doesn’t go away as a concept and as a governance tool. Fabric is the place where everything else Timothy lives, and I think what they’re doing is realizing that, like, it needs to have the ideas of lineage and and you know Billy lookup data and simplification for the power BI consumers right within that interface. Actually, if you’re power BI user, you just drop yourself into power BI right now. You see Fabric stuff like. That’s the thing. So cool about it. It’s not causing them to like even learn much of a new interface. It’s providing that right within the world that they already live, so there’s quite a bit happening and I’ll talk more about that in a second. So if you pivot into how Microsoft is positioning this, I think this is a great way to think about the four main pillars of how they think about Fabric. So one is assassin Fied analytics platform. Everything in one place for you to go and surface data that exists either completely in one link or data that exists in other sources that you’re attaching via shortcuts and doing so in a way that requires less data, data connectivity work or infrastructure management. And it just do not underestimate sometimes. Yeah. Do not underestimate the power of the emperor. Now, do not underestimate this state where you are. Previously needing to build up a complicated Azure architecture to get that data to power BI like the numbers of projects that we’ve had to do just to do that and then build up the appropriate or CAF landing zone for that to live in and secure it. And it’s just so much goes into that picture. The idea of moving that to the side and just connecting the data in is really exciting, so that is not that you’re not still building a pipeline and so on, but the fact that that you’re really focused on the pipeline and less on the infrastructure of the pipeline is something that’s really exciting to me. So the idea of a complete analytics platform talk about that more lake centric and open what that really mean by open is that it’s all parquet. It’s a platform that is following open standards. It’s not something where they’re trying to like, it’s interesting for them to say that, right, because they’re like, totally monopolizing the idea of the data platform. But it’s also trying to be open at the same time, so you can learn more about that picture. Umm. Ultimately, I think why the why of this is in a simplifying the data side over here and then even more over here, right? I think what the main results of this will be is like the shift that we had from SSRS reporting to Power BI and there was three stages. There was like the SSRS and then there was people like there’s all those like interim versions of Power BI that just never really made it. But then there’s like, there was that shift to it building power BI reports to then the democratization of power BI and it’s ability for any person to create meaningful views of data. I think we’ll see that same shift in the data engineering space, maybe to a slightly less degree because of the size of some of the problems that are being solved. But you already see that in the data science world where, like data scientists just want access to the data, they don’t want you to build a report for them. That same thing is going to happen with copilot on top of data, where you’re asking questions of data. You’re asking copilot to build you a power BI dashboard. Those kinds of experiences in the hands of real consumers is going to further change the game, and I think that’s truly what you’re unlocking is that ability to still the lock in the the version of the truth. But then be able to democratize access to data to gain insights that let people take action based upon that. So if we talk first about like what, what unified data platform means in the era of AI, it’s all these things in the context of a SAS platform. So data factory data movement still exists, but it’s done in a SAS applied way. Engineering science warehousing still exists, but in the context of assassinated way. Power BI already existed in that same interface, and then activator. Very interesting new technology that is available and then one lake being essentially the OneDrive of data storage and then essentially what you’re doing is you’re getting capacity against these different functions that you share across the estate. So you’re not like, what’s my charge for, synapse? What’s my charge for this? What’s my charge for that? And now all of different things. You’re simplified. Sort of purchasing process I think is going to be really helpful for many companies. So same idea around this idea of 1 OneDrive for data. It is a single SAS lake for the whole organization. It’s provisioned within the tenant. The second you start using fabric and it facilitates many of the mesh concepts via a security foundation using serverless compute, doing the same kinds of workloads that you we have wanted to do but done through having one copy so you can still separate the permissions, but you now have a separation between the functional areas and the storage areas in a way that allows for that import export to still happen but allows you also to enforce the security model across the pattern without having to say I’m going to take this Finance content. I’m going to create another copy of the Finance content because it’s their team wants to use it for a certain purpose. I can start to simplify that same picture, so really I find this just really exciting that they’re starting to go down that same kind of path. So, alright. Yeah, since of time I will just sort of move that forward. Ultimately, I think the idea of like having an RBAC model across something that once I put it into one lake, I replicate the data there, I make it available. I bless it with certain umm certified data set qualities. It then becomes something I can use security, then govern who can access and have less dispensary dispensing of the data across the estate that creates the same problem we have when people are using Excel in the 1st place. So there’s all sorts of interesting innovations happening. The one that I think is most interesting is this idea of Direct Lake mode. We don’t have a ton of time to explain it, but the sort of like basis of this, it’s I’m gonna speed up the ability for me to deliver more value by having less time to duplicate and having the ability for me to provide value faster. So definitely go research some of that if you want to learn more about the details of direct link mode. We don’t have a little time to kind of go into it right now, but I think it’s a really interesting example of an innovation that’s being delivered on top of what Fabric has the capability to provide, say with mirroring like I think mirroring is another really great example of that. Umm shortcuts. I mentioned this earlier. Shortcuts are the idea that data that exists in other places can be poked into fabric as a accessible data store. So if I have data already in Azure already like let’s say I have. Not Cosmos database. Probably bad example cause I can mirror that, but like a synapse warehouse or something in Amazon or Google like I have the ability to shortcut that data without data movement into one link and then make that available to other functions. And I think that’s great because it’s respecting the fact that there’s data across many clouds and is also consistent with that idea of a data mesh. So you want to learn more about that? I think that multi cloud picture is something that, you know Microsoft already kind of had with power BI, right? Like people are still using power BI as a primary consumption method. Now I’m sort of doubling down on that extent that even if you had data in these other locations, I still can make it a shortcut. So I can facilitate access to that and this place and that’s even happening in the context of dataverse. And that’s a new thing. Like I was actually kind of surprised initially when they didn’t have data versus connectivity into fabric. I’m like that’s a big limiter, man. Like a lot of stuff getting built in data verse, that’s now a thing that’s in preview as well. OK, I think the this idea of data domains has taken a huge step forward with data mesh and it was a concept that was kind of coming into the fold even a couple of years ago because people realized that this idea of throwing everything in a big snowflake instance was and that that’s the only place our data lives was never necessarily going to work because we still had data in all these other places. So, like we had data in this application and this application’s hub or this other place over here. So we needed to understand that the data was primarily going to be there, but we had to have a further understanding that we had to have separation of concerns across these different locations and how we actually go about moving it or not moving it. So what one lake is doing is it’s trying to facilitate this access to data without having to do the data movement and facilitating essentially RBAC based access to these different sets of data via these sets of concerns. And then can be certified by the domain experts. So, like Finance could say this, this and this is a certified part of the data set that’s that exposed to other parts of the ecosystem without having to like just copy the data or have them have complete ownership over all parts of the movement. They can have a very business centric understanding of the data and certify that part of the lineage without having to have some of the complexities that we had in place with other data mesh concepts. So I think understanding how one lake allows data mesh concepts to be brought into real life without having to have as much data movement I think is a really nice win and especially simplifies the picture for a lot of companies. OK, so this idea of empowering every user power BI obviously doing work in the context of like data science and so on. Clearly, something that they’re launching in in one lake is something called data activator, which is a new code experience that allows you to trigger different rules based upon things happening in the context of the data. So if you this very similar to like. Some things you might have done like IT hub and so on, like the ability to see data, data flowing, be able to trigger actions based upon that data and be able to either process in a certain way or fire off an email or do something that back up here. So that idea is something that because they owned like the other parts of this action platform power automate, I could see copilot studio coming into play here. Teams. Other custom activation on data, I think that’s gonna further enable what people do with it. Like I think another way to say this is many companies started with having power BI be the main like reports, right are the main way we consume data and we take that we take action on it. People are now discovering, like, it’s not just reports are just a means to an end. Like I’m actually trying to do a prescription. Uh, I’m trying to predict and prescribe action that is being taken based on data like and that’s something that a human oftentimes does, but maybe I’m replacing that with the data science workload or something else that’s happening. And if you think about it that way, but then as you get in this picture of even when I’m predicting and prescribing, maybe I’m actually doing something with it, because once I’ve prescribed, I can just take action and that’s where the sort of world of AI agents comes into play that we’re not just going to be like, waiting to consume something in a person looks at it and does something we’re going to say, I know that when this happens, this other thing needs to happen. And I can build an AI agent that takes action based upon those things. So integrating into things like power automate and copilot studio and teams are going to be opportunities for that action to actually happen without necessarily the human involvement on every step. So data activator is really cool. It’s kind of leads us into this idea of AI powered are kinds of ideas, which is a whole conversation in and of itself, but one that is really exciting, like many of the conversations I’ve had about Fabric, have been entirely spurred on by people wanting to use copilot in power BI like the idea. Like I want to use copilot and Power BI on my data and I love the fabric is built on the idea of that. You should know that if you’re going to use copilot on your data, the big feedback from the people doing this actual work is you need to structure your data well and in Fabric for it to be useful by power. Uh. By copilot like. Yeah, but like, it’s definitely something that we always think about, right. So the idea of labeling understandable data structures, being able to surface it for the right users, being able to do that in a way which can be consumed in a coherent way, that’s all copilot accelerated for. Sure. Because like once that pre work is done and lineage is in place, then a copilot can build the report for the person they need or answer questions about data that you would be asking about. Or you might even integrate a custom GPT. Been building in copilot studio right into the Fabric interface that I think is really where the magic is going to happen. Like this idea of all this commoditized building on top of the data and top of fabric. Amazing. I just. I’m I’m so excited about the future of that versus like it being just sort of walled off inside of an IT organization. OK, so how did this work? From like the the number one question I guess like So what does this cost like? How do I even do this? Uh, so it’s kind of broken down into two things. It’s compute capacities and storage essentially, so you get a little bit of that when you upgrade to power BI premium. There is a sort of capacity that you buy as you go through acquiring it. And umm, I’ll skip over this. I think this is, yeah. Where did we go here? I’ll get to that a second actually. So OK, so I’m gonna. I thought my capacity section was coming up. It will come up in a little moment here. So capacities you buy them and you buy in just like sets of stuff and that’s shared across the entire set. You could say that’s good or bad, but that’s essentially how it works. You don’t dispense each of these are not bought in and of themselves. You buy like a CU and then that CU is used across all the functions in compute capacities and then you have storage capacities at the end of this presentation I have some kind of breakdown of what that essentially costs, so I’ll show you that in a second. So there is a process of moving to Fabric from other things. If you are using synapse, Umm synapse in any capacity, I think it is really important that you think about how Fabric will be the future of what you are doing. It is a much easier way to do a lot of things that they’re building. The Azure synapse experience or Azure data factory or the Azure Data Explorer, but they have made it easier to use those same things in conjunction with Microsoft Fabric. So you can do a little research on that. It’s definitely something they built pretty well. So from a data platform standpoint, we talked about 1 Lake, we talked about 1 Lake has ability to surface data through different security layers. I think something important to comment on a little bit more is how Microsoft purview fit into this. I mentioned at the beginning that Microsoft purview initially wasn’t very well integrated into fabric. That has changed in the last year. They’ve done the right work to be able to fit those two things together. Purple is a little bit of a question mark for a lot of people because they took per view, which was very focused on the data state and then they renamed the Office 365 side also to preview, to call it all Microsoft previews to be called Azure purview and it’s called Microsoft Purview. So to make sure there’s no confusion, there is the Microsoft purview capabilities for Office 365 centric data files on teams and so on. And then there’s also the Microsoft purview that serves primarily the data a state Big D Data Estate. And theoretically, they’re the same product. In practice, they’re kind of two sides of a coin in a sense of like their ability to engage across those two repositories. Like, anticipate them coming together. But in this case I’m talking more about the data estate side purview, so its primary functions in this ecosystem are the one link data hub, the ability to create endorsements or certification of data via certified data sets. It’s ability to create data lineage and impact analysis associated with certain data that’s similar to the same data lineage that exists in fabric. So I think that’s something you’ve seen happen that didn’t really exist a year ago. Is that kind of uniting of those two things and then the ability to extract data and look for sensitivity information across data state and be able to label and know where sensitive data exists? And that’s something that many companies are really looking for. I was talking to a company the other day and they’re one of the biggest needs was I don’t know, word sensitive data lives across my state, and it’s even more complicated because I have data in Snowflake and this other place and this other place. But I don’t know where else it lives in my cloud environment and it’s like, well, you’re covering like 25% of the data, like there’s all these other apps that have data Cosmos database or this operational thing here, this thing here and they all have data that’s important. So Fabric is very close to 1 to. Sorry, purview is very close to Fabric, but purview is also broader than Fabric because it realizes that like data exists in a lot of different places operationally and covers that as well. That’s why there’s still two tools, in a sense of it’s general data governance. So there’s the one that did hub. Its main goal is to facilitate discovery. Umm, you know, reuse finding data. It’s more of a user experience, consumer user experience. There is the idea of endorsements. So the idea of surfacing data specifically for the purpose of saying this data is one that you should use. It’s a certified data set. This is like you’re a consumer of data. You’re a data scientist or a consumer. Use this data over this data that’s named the same way. That’s something we have a lot, right? We have these different examples of data, but we don’t know which is the one that we should actually be using that labeling of data. That creation of catalog, the creation of information that the data should be promoted, is something that now is right within the power BI experience because it was separated right like you had it like in purview, but not in the power BI experience. Now that’s part of part of what they have. Fabric. The idea of data lineage, I think this is super cool. Like the relationship between where it came from and how it got to you. And then the idea of understanding its impact if you were to adjust a certain data pipeline, all happening within the same experience as Power BI is a huge step forward like not having to. Here’s my power back experience, but now I want to see the lineage, so I’m going to go over here, like being able to do that all in the same context is beautiful. And then also is surfacing only data that people should see. Maybe like I have a lineage, but only goes back to the spot or only shows me these trees because realizing that I only have access to certain parts of it I think is a really good part of this as well. And then the ability to understand what data is where based upon the ability to scan that data and see where sensitive data exists within the ecosystem, something that per view also brings into the same picture. So how this works from CU standpoint? We talked about this just a second ago when you buy cuz they’re all pooled together, you don’t have to preallocate, you can buy it as you need it. So like if you just, if you’re like a low capacity user and all of a sudden you need a lot more, you can allow it to scale up just for that use case. And then scale back down as you don’t need it, which can be really advantageous activity. One of the things I did include in this is how fast some of this stuff can happen when you’re using connected data sources. Microsoft already done the work on the ability for data transfer to occur substantially more performant than what it was even able to do in synapse, or beat the pants off Snowflake. Kind of stuff is just really crushing it really interesting. I haven’t put much of it into this deck, but I think it’s interesting to look into if you wanna kind of get geeky there. Umm how this works for a pricing standpoint? Essentially, there’s a skew, a capacity unit, and you can pay for that capacity on pay as you go. Or you can start to reserve my recommendation or just start going pay as you go, or just reserve at the minimum level and then determine what your actual consumption on a per monthly basis is. I know you look at this and you’re kind of like that’s kind of French to me and honestly, it’s kind of French to a lot to everyone because we’re still kind of figuring out what capacity truly looks like in fabric. And that’s just on a statement on that. Umm, because it is something that’s still like emerging, but I like the fact that you’re basically just building buying capacity and the more capacity you buy, the cheaper it gets per thing and you start to then do reservations, which was really complicated in Azure, especially with all the diverse spots. And then you can start to simply save quite a bit once you know what you’re going to use. So like once you know you need a certain amount of capacity, you can start reserving very quickly to be able to get that capacity. Uh, same thing with one lake. So 1 Lake storage is based on pay as you go you can start to go down the reservation route, but you can see the page you go price of 1 Lake storage there as well. And then there’s a lot of question marks around the data transfer, Internet egress kind of things. There is a bandwidth pricing space and we can provide a link to that and know if you can actually click on this deck. Probably not. Yeah, I could provide that link in the chat. Umm so that. Brian Haydin 52:36 I think Nathan. Ohh, have they turned that on yet? I think that they they haven’t turned on that pricing for egress yet. Nathan Lasnoski 52:44 Oh, thank you. Brian Haydin 52:46 Yeah, I I think that’s something that they’ve. They’ve warned that there’s going to be, uh, but but they haven’t turned it on. Nathan Lasnoski 52:54 Thank you for that introduction. I didn’t realize that wasn’t turned on yet, so I guess now is a good time. Brian Haydin 52:58 Or or maybe it is? Nathan Lasnoski 53:00 Let’s check so we can validate that. OK. Holy cow. So we actually got through that. So now is a good time to pause. We have 7 minutes. We definitely did not cover everything there is to know about fabric and we don’t know everything. There is to know about Fabric, but we’ve used it now, so we know quite a bit now. It’s a good time to ask some more questions, so if you wanna drop some more questions and then chat or you want to ask them if either Brian, I’ve totally been apologized. Brian, in this conversation, he’s actually way smarter than me. So if you want to ask some more questions, now is a great time for those questions. While you’re thinking about that, I had a question that was asked on the side, which is I’m building a like less consumer oriented platform or sorry like a less data analytics internal oriented platform. I’m building an external oriented platform. Should I still put it in fabric? I have a I have an answer to that, Brian, I’d love to hear you. Like maybe expound upon that idea a little bit. Brian Haydin 54:27 So from the beginning it from the beginning of Fabric last year when it was released, they they did enable API access to your data. So that is something that is part of the roadmap to be able to operationalize that data, but it is an analytics platform first. So I think that over time, you’re gonna see uh, those features. Uh, you know, become better and more usable for like applications or operational datas, data needs and but it is possible, but it’s just getting better. So that’s my perspective. What’s yours? Nathan Lasnoski 55:08 Yeah, similar like it’s clear. It’s like built internal analytics, first with like data science, but I didn’t cover the data science thing. There’s like a whole persona map of different roles and optimized experiences for roles, and that data science, exploration spaces, pretty new in there, but like very cool. Umm, I think that’s gonna start to take hold that it’s it’s also the place for data that might be used in a like external operational platform. But I think that’s something that will emerge more. Very interesting. OK. Other questions. If you ever come off mute or you can drop them in the chat either way. OK. All right, so before you leave fill out the survey in that survey, I would love to hear feedback on whether this was helpful. This was or not helpful or you loved it or you hated it. Either way, I always love feedback. I want to know what worked and what didn’t work. Hopefully it hit all the right nerves in terms of information you want to learn from this, or at least was a good start for you, so that is something I would love for you to do. It also includes a potential action item of you having an envisioning workshop to learn more about your data and how Fabric might fit into it. And that’s something that myself and Brian would love to have a conversation with you about. So as you leave in, that would be. If that’s interesting to you, please push that button and we would love to get some free time with you to talk about it. So other than that, if there aren’t more questions, I wish you all a wonderful day and a great first day of March Madness. I’m sure that many of you will enjoy that and a great start of spring. Thank you, arca. Brian Haydin 57:05 Go Marquette.