Revolutionize Data Analytics Strategy With Microsoft Fabric

First, thank you to all of you who are choosing to spend some time with us today to learn more about Fabric. I think it's a really, really cool, really interesting topic. But I know there are a lot of ways that you can choose to spend your lunch hour. So really very much appreciative that you chose to spend it with us in Insight. Just, I guess, a little bit to kick us off here. When I started with Microsoft, gosh, 13 years ago, I was leading a team of sales and technologists. It was called Application Platform. And it was all about SQL, BI, big data, some analytics stuff. And I just think about all the things, and it was pretty new to me at the time, all the things I learned at that point about what you do with data, how you capture it, how you get it all structured and in the right place and in a single large box so that you can leverage some analytics capabilities against that. And then sort of fast forward to, I guess, at some point this late summer, early fall, the first time that I really got a deeper technical briefing on Fabric. And look, there's certainly been incremental progress along the way. But I sort of went back to that 13-year-ago version of myself and thought about, what would I think of Fabric if somebody popped that in front of me 13 years ago? And I would have said, it's magic. It's impossible. There's no way in heck that we can make things work like that. So I really do. It's probably the single with all of the announcements that Microsoft has made, and we've made a whole bunch over the last six months or so. I think Fabric was the one that really arguably got me the most excited, particularly as a business user and someone who needs analytics coming from tons of different sources and needs them available immediately. Seeing the power of Fabric, I was just blown away by it. And I hope with everything you learned today, you sort of have a similar feeling. So, look, we announced at Ignite, gosh, was that last week, week before, general availability of Fabric. So it is out in the wild at this point as sort of an all-in-one analytics solution, designed to bring the data estate into the AI era, or I should say the generative AI era. It does a lot of cool things, integrating Power BI, Data Factory, next-gen of Synapse, and kind of doing it all in a unified, simplified manner, and doing it at, most importantly, a price point that is manageable and is not going to break budgets. So just, I guess, a quick high-level teaser about the things that it can do. So real-time data processing, but real-time data processing at scale. And I think that's something that is just absolutely critical today. Data integration, to me, is probably the most amazing part of this. So sort of synthesizing all of these data sources and pulling it into just like a cohesive single analytics framework, I think is something that is probably more than evolutionary, bordering on revolutionary. The collaboration, right? So it enhances collaboration across teams. It's a unified platform for data analytics and can really be leveraged across all kinds of stakeholders, right? All levels of an organization, all departments of an organization. And then at Microsoft, security is something that is absolutely critical and essential. It's in our DNA. So having security sort of prioritized, built into Fabric from the very beginning to deal with all of the global regulatory landscape, compliance, and just making sure that the most sensitive data can be protected. The next thing is AI integration. Obviously, AI, hopefully all of you believe, it's more than just a buzzword. AI has been out there for a long time. Generative AI is a bit newer concept to be out in the wild, but using the predictive modeling to generate those automated insights is something that is very new and very exciting. And, you know, I guess I'll end with this, right? Insight is a partner that I've personally worked with probably my entire 13 years at Microsoft. And really, look, they stay on the cutting edge with us. And that's what's most important. I think there's a lot of wonderful partners out there that can come talk and help with things like Fabric. But the thing that I love about Insight is just their focus on technology, right? And how on the cutting edge that they stay and learn with us or engage with engineering and really are just absolutely fabulous partner, particularly when you're going after some big, heavy, really meaningful things that are right on the edge of the emerging technologies. So, again, just a huge thank you to everybody for being here today. I'm the, I think, the last talking head that you'll hear from. So I'll turn it over to Brian and we'll get into the meat of it. Thank you, Ben. Can everyone hear me okay? Yes. Yep. All right. Let me jump in. It's a little about myself. One thing I do, I have a YouTube channel. And you can see, I don't know if you can see it, but I'm branded with my Spark mug. And I'm also wearing my Microsoft Power BI polo shirt, which I usually wear on my channel. That one I have in the picture is a Python shirt. But I like to wear something in the theme. And you're going to see that Spark and Power BI are definitely two key parts of talking about Fabric. I'm a data architect consultant at Insight. I wrote a book. I got so excited about big data and AI and everything that I wrote a book called Master Azure Databricks Step-by-Step. It's riveting. I also wrote a book a few years earlier than that called Pro Power Shelf for Database Developers. I've written courses on various things, including R. And I've probably worked in just about every area of the industries over the years. So I've been into everything. I was previously at MVP, Microsoft MVP. I actually worked at Microsoft a couple of years. And I do have actually an MBA concentration in finance. If you've heard of a school called Bentley University, that's my alma mater. And let me jump in. So overall, what we're going to be talking about today is insights and AI on a modern data platform. I'm going to talk about how Microsoft Fabric fits into that and encapsulates really all the requirements that we're going to be discussing. And Ben gave you a very high level summary of what Fabric is about. I'll delve into it a bit more to show you how it looks and feels. And then we'll talk about how we engage, how we can engage with insight to help you jumpstart your initiative into this brave new world, which is exciting. But, of course, you have to have a lot of knowledge to do it. So we're going to talk about really is the state of the AI marketplace, which, as we know, is very dynamic. Getting your data estate in order, driving ROI with your data investment, maturing analytics capabilities, getting that solid foundation that you can build on and go forward. That's crucial. That's where insight really steps in. Core companies that insight offers, top challenges that we see with our clients. And then, of course, Fabric and how that all ties in and how we can help you get started with Fabric and not just get started, but really lay a foundation. So this slide tells a lot, but it doesn't say anything other than what I already know. The different types of data, the volume, the different architectures are resulting in companies being inundated with all different types of data. It's actually overwhelming. And what I keep seeing with clients is they just don't know how to get a handle on it. They're just trying to nail down one data source when something new comes in and then another and another. And that's aggravated. And that's the left bullet, 72 percent. The 62 percent slide tells me what I also already know, which is the organizations are kind of on a legacy architecture. They're siloed. You've got accounting, you've got finance, you've got operations, marketing. And while they work well within themselves, they're not really designed organizationally or technically to integrate and to be able to really leverage across. And when you get into the AI space, it's critical that you break down the silos and make sure everything's working seamlessly or you're not going to be able to get maximum value. And on the right side, I think 50 percent is really higher than I would expect. I have not worked in a place yet that really is fully using their data. But to even approach getting it to 100 percent, you need help. It needs a lot of expertise and a lot of deep dive knowledge to take you from zero to hero so you can start leveraging your data, which is now not even a question. If you don't do it, then basically your competitors will and you'll be in a bad position. This is what is happening right now. Companies are embarking on their AI journey. Chats GPT has really hyped this up. But even before that, there's been a lot of movement. I think any company right now realizes they have to get a handle on AI. How can they use it? It can improve productivity. It can produce entirely new products that they hadn't even imagined before. It can reduce costs in any number of ways. But they have to start getting a handle on how are they going to leverage it. And the key to leveraging AI, actually a standard metric that I've used again and again, is any AI training pipeline is built on a data platform, a modern data platform. 80 percent of the work is actually the data engineering, getting the data from the sources and into a form that's usable for data science and analytics. So that gets to that sort of below point challenging supplying data science teams with data. That's a huge problem because of silos, but also because the data isn't formatted correctly. It's on prem. It's not in the cloud. It's not in a place that can be easily accessed. And that's aggravated by all the changes going on in the industries. People are being acquired. People are splitting parts off. So that's another part that just increases the need to get a handle on your data estate and start building successful ROI projects on top of that. On the right in this slide, you see business intelligence and AI, which are clearly extremely important. BI has been around a while, but still not achieving its full value. And the AI is sort of the cherry on the cake. It's a critical one, but it's really not the cherry anymore. It's becoming so critical that it's becoming the most important element because now we've got good AI models that can do things that we never thought were possible. None of this is possible, again, without going back to the data platform. And that typically is going to be on a cloud like Azure with infrastructure being built out because of really the flexibility and extensibility and scalability you get there. And of course, security and governance is crucial, or you're really dead in the water. Even if you've got the most beautiful AI, if somebody gets a breach, you can't allow that. You'll have problems. So you have to make sure you have good governance and security. So what we're really trying to talk about here is in the lower left-hand quadrant, this is like your typical Gartner quadrant kind of thing. Most companies are in the lower left. They've got low business value, data siloed. It's not formatted correctly. Some of it isn't even useful anymore, but it's being used. And everybody's pointing fingers like, which data is correct? What companies are trying to do, where they need to go, and what Insight does is bring you into the upper right quadrant. You're getting strategic value out of your data. It's delivering mission-critical applications and models so that you are now a step ahead, or hopefully several steps ahead of your competitors, and you're going to get a lot of cost savings and profits. One of the big things I see a lot is companies turning it around so they provide a service, maybe like analytics dashboards to their customers, or better, AI models with the dashboards so that they can get business value for themselves, and that's monetizable by the organization that creates it. And a kind of point to this slide really is it requires teamwork, because when you look at this typical data science flow, it starts usually with the business. Hopefully, if you're there and your manager is saying, where should I start? What do I do? You start getting your head around that, bring in an expert consultant firm like us, I guess, and say, what should we do? That starts the initiative, but then you bring in the business analysts and the data analysts, and they'll start looking at the data. Then it needs to be prepared and called out from different sources, and that's where the data engineering kicks in. The data scientists will take over from there doing the model training. Then it goes back into the business, integrated. But the thing that people don't realize is it's not like a one-and-done solution. There's something called model drift, which means the data changes, the underlying business changes, the environment changes, so a model that was doing something well even a week ago can be completely invalid now. So you have this cycle going around and recycling, and of course, you're going to be always looking for new opportunities. The takeaway is you need to get a lot of collaboration. Everybody needs to work together without having these barriers. So the number one challenge across the board globally, I believe, is harnessing the power of data, and that means you need to create a unified data analytics platform. If you can get those two things in place, then you're in a good position to have a competitive advantage with AI and analytics on top of that. And at that point, you're in a good place. You just got to remember to keep in your scalability and security. You can't sacrifice that at any point. But when you get to the third bullet, you're at a point where you are going to be beating your competitors, and that is the new competition in that particular space. So this is sort of the how does Insight help you with this slide. And the thing is, if you're like most companies, you're probably struggling. Like, do I really have to worry about chat GBT? I've heard problems with the open AI being safe and secure and not revealing data. How do I leverage this for my company? We have a lot of problems. We're an old firm. We don't have the right things. We don't have the right skills. So the beginning point is management realizing they have a problem and reaching out. From that point, we can help you. We can analyze, we can do research, and we can define where you are currently, and then we can go into the envisioning step where we say this is where you could be. And you can obviously help with that and say we want to be here. We want to be number one in sales. We want to be able to turn products around faster. The next step, which is really crucial and sometimes companies leave out, is making sure sanity checking, as I call the proof step, making sure that it will work, that you've got the right approach. And so you use proof of concepts, you use minimal viable products, things like that, that will start off something and say we've got something worthwhile, but we've also proven that this is doable and we're on track. And then you can start the acceleration process. You put that in and you expand on it, you scale it, and you go back in the beginning and add more models, more analytics on top of that. And it's all going to have to be kind of managed in some sort of an ops process. And we've got tools. You'll see something called MLOps specifically designed to do that for the AI stack. So I'll give you an example of how Insight has helped customers. We had a client with over 100,000 different components, and their customers of this particular client had to create different custom assemblies of those products to deliver. The problem was, and this is very common, they did things in a very old-world way. Everything was manual. People had to get involved at every step. Certain expertise was required. So it was very time-consuming, and that drove the cost of the product up. So Insight came in, and they searched out where the data was, how to bring it together. They pulled it from the CRM and brought it into a Databricks lakehouse. And from there, they were able to speed that up with an Insight proprietary metadata-driven framework, which I cannot say 10 times fast. That allowed them to get a data lakehouse built within days as opposed to months or years, which is a huge challenge. So that's kind of the value add we were able to bring into that. We were able to get a proof of concept built, build a model, integrate that with Power BI, and then kind of get into a testing loop where we could show the business, this will work. This is the metrics we're seeing. We can definitely get this kind of return. So once that was implemented, within eight weeks, the company was now getting $15 million increased revenue a month with a projected annual increase of $50 million. But that was just like one model. If they go back and do the same thing in other areas, maybe with inventory, maybe with sales costs, any number of areas, they could then keep recycling and expand on that. And that's why this is such a competitive advantage. And as I mentioned, a lot of times outcomes can be tied to monetizing, creating entirely new products based on AI. So now we talk about Fabric. What is Fabric? In a nutshell, it's a complete analytics platform. And I'll demonstrate a little bit about that so you can see it. But everything is under one user interface. And the tools are designed to customize to the perspective of the user. So what do I mean by that? What I mean is you've got data engineers that can work in the same space with data analysts, with business users, with data scientists, with Power BI report developers, and anyone who needs to access data to use it in a secured and governed way can go in and do that. We'll hear about something called OneLake, which is a crucial part of that platform. It's lake-centric and open. And that means it is OneLake, as I mentioned. And that, we're going to talk about, allows us to have a single pane of glass into our data. What's great, too, is everybody knows Office, right? You've got all these Office products, including Power BI. But integrating them can be a challenge. But Fabric allows you to do all kinds of things, including something which I'll show you is actionable insights, which allows you to trap certain events that are crucial to the business and surface them through notifications or emails or different types of Office products, or present visualizations and analytics through, say, Teams. And of course, Microsoft has been driving the entire AI revolution we've been seeing over the last year. And CoPilot is their branded name, basically, with their intelligent solutions to help everybody in the organization be more productive, build solutions more quickly, and reduce costs and increase revenue. So they also use ChatGPT kind of under the covers for that. But all of it is secured, so you don't have to worry about, is anything being exposed? And we have a lot of abilities within that. So these are the kind of core elements that we're talking about. So I want to step back a minute, because there's sort of two concepts I want to explain here, because it can be a little confusing if you're not used to Fabric. The core thing when we talk about Fabric is embedding something called One Lake. But let me start with Fabric. So in a nutshell, Fabric is a comprehensive software-as-a-service user interface to do all of your analytics work. And it integrates really many tools, but it calls experiences things like the Power BI experience, the Synapse experience, Data Factory, and I'll talk about that in a minute. But it's kind of like a perspective you take on when you go into Fabric that fits what you want to do and how you see the world. It's a self-service, all-in-one analytics platform. Now, a key element that you cannot overemphasize is One Lake. One Lake is really, I think, a game changer, because One Lake does something that we've lacked in the past. As long as I can remember being in BI, there's always been this holy grail where you have one source of truth, one enterprise, whether it's a data warehouse, enterprise lake, that everybody can share and use. But it's always been very elusive because of all the fragmentation technically in the world. How do we get things together and get people to work together? What One Lake does is it basically creates a logical layer that's so abstracted, you don't even have to worry about storage. You don't have to allocate things and say what you want to turn on or off. One Lake takes care of that. You just worry about, I want to have my data exposed. And the key thing is you can do that with a consolidated view. But a crucial element to this is you control what's exposed in One Lake. It's not like everybody can just see anything. And that's important because you could have data that you really don't want people using. Great, don't put it into One Lake. Or if it kind of gets outdated or you've replaced it, you can just remove it from One Lake. So you always have sort of your pristine, where the current good data is that the business should be using. Now, I want to carefully explain something because a lot of times when you tailor One Lake, they'll compare it to OneDrive. But OneDrive is, in my mind, I think I have my own little personal OneDrive. That's not what One Lake is. One Lake is truly One Lake per tenant. And so that means you have One Lake for the entire organization, but everyone can see into it. And you can create, as we'll see, a workspace that kind of gives you a view of what you want to see out of the One Lake. But it is really nice because you've got this sort of one giant lake that your organization can see and only exposes what you want. And of course, there's data governance and integration tools that you can use around that to maximize the value. So kind of a snapshot of Fabric. At the core, the platform underneath. So whenever you write data out in Fabric, you'll be writing it to One Lake. And the core of One Lake is using Delta Lake. Now, what's a big game changer here is that in the past, you might think of proprietary formats and things. But Apache Spark came up with a format called Delta. And Delta has become... Now, the key thing is it's transactional, so it supports updates, changes, and deletes. So that was the key difference between, say, Delta and Parquet. So Delta allows us to create essentially a data warehouse on a data lake. Hence, we get the word lake house, right? That's how they kind of munch the words together. You have updatable data in Delta tables, and you can think of those and use them just like you would SQL tables. Slice and dice and all that good stuff. But it's infinitely scalable because it can use Apache Spark or any kind of scaled-up platform under the covers. And it's open source. Anybody who...almost any tool can read Delta tables. So that's kind of the core underneath. And I want to show out in the corner, you can see flat files because the idea in OneLake is it can also support linking in. And that's what they call it. You create a link, and you can link in, like, a flat file folder or blob folders or other things that you want to add into OneLake so that it can be shared. That can be brought in as well. Now, we talk about Fabric itself. I want to kind of reinforce a few concepts. One of them is this idea of experiences I mentioned. Now, we've had Synapse around for a long time now. And people like that. The good thing about Synapse is it really is SQL Server scaled out over a big data platform. So you can invoke it and have, you know, petabytes of data. That's awesome. It's a good thing when you have people who are coming from on-prem, you know, SQL Server, and they're comfortable with using SQL tables. They like stored procedures. They like to have that grant permissions kind of feel. They want to use T-SQL. Well, you can do that using the data warehouse experience within Fabric. And the nice thing that's going on here is it's still writing out to Delta Lake, so it's not locked in a place where you can't get it out with open-source tools. But say you actually like Spark and you know Spark. That's kind of where I'm at. So I use Spark all the time, and I use Python a lot, but I also use Spark SQL. And Spark SQL is not the same as T-SQL. It has a different feel to it. So I may prefer using the data engineering experience, which is essentially a focus on using Apache Spark. And you've got notebooks, and you've got tools, and you can schedule jobs in there as well. So if you prefer to work with that experience, you're open to do that. And we can collaborate together because, again, it's all going to the one lake. Another place I would typically kind of gravitate towards is Azure Data Factory, which they shortened to Data Factory within Fabric. Now, Data Factory is kind of like does everything, right? It's your Swiss Army knife. So if you can't do something, the other tools, or you just want end-to-end everything, then you can plug in Data Factory and create pipelines that go from any source, transform, yada, yada, into an endpoint. And you can now make that available to the business to use all in one lake. So that's kind of the person like the data engineer might use. And data scientists probably might be more in the sort of middle tier there in some ways for what they're doing. But it's all a matter of what they need and what you're comfortable with, really. Now, all of this is to build data, data pipelines, right, and get data into a place and format it's needed. Once you have that, data scientists can use the data science experience where they can access open AI models, lots of open source frameworks like PyTorch, on and on. Or they can use pre-trained models using Azure Cognitive Services, which really covers a lot of more common use cases, and you can get up to speed instantly. Now, a crucial element in this is it supports something called MLOps. MLOps, using something, again, another term, but MLflow. MLflow is an MLOps open source solution. It's supported by Databricks. It's supported by Apache Spark. It allows you to get the DevOps experience, the DevOps functionality in AI. And that's crucial because when you're training models and you're putting it through all these paces, you need to track. You need to save artifacts, like save the model. Save the model from last week. Add versioning on my models. Add versioning on my data that I'm putting through the models. What did it look like? If something goes wrong, I need to go back and say, what was in the data? What outliers may have thrown off my model? So all of this has to be tracked, and that's what MLOps does, and it's integrated. Real-time analytics is the sort of Holy Grail, also, of things now. Everybody wants their data yesterday. They want it before it even comes in, and that's streaming, right? So using Event Hub and real-time analytics in Fabric, you can stream things in and capture it in real-time, visualize it, whatever you want to do. But what's really cool, I think, on top of that, is using something called Data Activator. So what's Data Activator, Brian? Data Activator says, okay, I've got all this data coming in, but there are some events that are so crucial, I need to act immediately. Imagine you're a store that sells lots of products, and sometimes you have a safety recall, and the safety recall could be like something that, you know, a choking hazard for children. Well, obviously, the liability, the risk, and certainly regulatory response requirements require you to jump on this, like, before it happens, if you could, right? You want to know and act right away. And Data Activator lets you build a filtering set and a rule engine that says, if I see this coming in my stream, I want you to do something, like mail the head of safety or security or product so they do something right away. And that's being integrated as we speak with more and more tools within the Azure stack. And Power BI is the sort of workspace in which Fabric has integrated itself with so that you can really maximize all of your Power BI analytics needs. But the takeaway that's really important from Fabric is that instead of the traditional ingest your data into a Power BI tabular model, which is a standard thing, what you end up doing there is you're kind of wasting time ingesting it, and then it's out of date immediately. Direct Lake is a new feature where Power BI just goes against Delta Lake as if it was an internal data store, slices and dices, and it does it really quickly. So you're not waiting for those clicks. You need fast responses, and it can deliver that. So that's what's going on with Fabric and all the different services underneath the covers. Now, as I see, I popped up Copilot because Copilot is the AI service that Microsoft is integrating basically across all of their tools. So if you don't know how to do something, Copilot can help. If you want it to write things, Copilot can do that too. Copilot can help you build your solutions and really bring you into highly being productive as opposed to sort of squandering time trying to figure out what you're doing. So one last thing here I want to talk about a little more about OneLake. So what's the big game changer with OneLake beyond what I've talked about? Well, imagine, you know, you've got OneLake here, and you've got something in finance. So you've got a Databricks pipeline that's been there for a year, and they've got this nice little data mart that's specific for finance, but nobody else can benefit from it. Well, you go into OneLake, and you can, from Fabric, create a shortcut, and you can add it into your OneLake. And now all that data in finance is exposed to the corporation, and you can leverage it. You've got someone else, something else here. The sales area decided they were going to build pipelines with Databricks. They're going to store it as parquet files. They can't, they're just appending data. But they decided that when they build the pipelines, they're going to write it directly to OneLake. And that's possible. You can actually treat OneLake just like a data store. And they decided to write their sales data directly to OneLake. So it's available immediately to everyone. Operations, they're using Synapse. Maybe it's the more legacy version of Synapse, or they're doing it in Fabric. But they can write also to OneLake and make sure everything's available. Accounting's sitting there saying, okay, great, but we've got this blob storage where we created a file folder that's got invoices and a chart of accounts and all kinds of reference data, budgets, whatever. We'd like to be able to expose that to the rest of the company. And you can do that now by, again, creating a short link, and it will point to that, and now it's visible in OneLake. And a really common thing I've seen is organizations are becoming very integrated. You have partners that you share data, you collaborate with, or maybe you have a vendor that provides crucial services where you want to get the data back and be able to view it. Well, you can set up a link also, and even if the data with your partner is in AWS, you can set up a link to it, a shortcut, that will bring it in, and you don't have to move the data. You can slice and dice and work with the data right where it is. It'll automatically get the data and bring it in when you need it, but you don't have to copy it into Azure. And something I forgot to mention here too, but a cool thing about OneLake is with a very quick little extension, you can install File Explorer on your good old Windows desktop, can give you an opening just like another folder, and you can see what's going on in OneLake. So you can see all the assets in OneLake through File Explorer. And I'll just show you this real quick. That's what you see on the right. So I installed it on my own machine, and that's a snapshot of my actual machine. OneLake shows up, and I can browse and look at my data lake, but really the whole OneLake through that. Okay, so I talked about this. I'm going to step right now in and do kind of a walkthrough, a quick overview of what things look like in Fabric. So when you first go into Fabric, what you see is the Power BI workspace. But if it's Fabric enabled, you'll know because if you go to the lower left corner where you see the Power BI icon, click on it, you'll see these new options. And these are what I talked about before, the experiences. So for instance, if you really want data factory, you want to do a more, you know, professional data engineer experience, you can go to data factory. And typically up in the upper left-hand corner, it provides you with the tools to create right away. You'll see also other tools listed below. So if I want to create a new data pipeline in Azure Data Factory, I can click there. It also added, in this case, Power Apps Dataflow. So if I want to use a more Power BI experience, I'm able to do that right here. What I can do also though, I'm going to just show you what does a pipeline look like. When you open things up, one feature that's kind of nice, kind of like tabs in a browser, you get shortcuts in sort of a task pane on the left. So I can bring up one of the pipelines I created, and you can see this was just something that a wizard walked me through when I was able to create a copy. They might say, don't I not want to copy things? Well, in this case, these were files I uploaded into One Lake, but I don't want to work with, you know, CSV or Excel files. I need to convert them into a Delta file format. So I tell it where it is here. I source, and you can see like worldwide importers, and then I tell it where I want it to put it, which you can see here is my lake house, and then where it's going, and I just run it, and it will do that. And I can schedule this to refresh however I want. You see the scheduler, run, validate. So I'm having a data factory experience right within Fabric. So let's go back here now and say, okay, but I want to get more of a data engineering experience. So that's Apache Spark. When you see data engineering experience, think Apache Spark. So you can see here, what are they exposing at the new point? I want to create a lake house. That's basically Delta tables under the cover. I want to create a notebook. A new feature now they're allowing is I can create shared things like maybe Python libraries. If I want to create a job, something I can schedule, I can use this. A data pipeline, some other things are exposed just for ease of reference. I can import a notebook, so I can go over here and say upload, and then I can just bring something in. So I've got all of that ability, but if I'm in this type of space, I can go and look at my lake house. Let me close that for a minute. If you see the icon that I opened, it's got like a house with like the waves over it. That will bring me into a view, which is a lake house view. And a really cool thing that you don't typically see in Apache Spark, of course, is I've got this intelligent grid that I can look at things through and sort, and I can kind of get a view of my data. So it's great for analysis. My flat files that I might be importing are here. So these are things I uploaded, and you can see I converted them. I can look at the data types and then the column names in there. So I can do a lot of work there, but I'm probably going to want to create a notebook. So let me go to here, and this would be an example of a notebook. So one of the things with notebooks, if you haven't used them, is that notebooks do not discard the output they create. This is creating some optimization things that Microsoft recommends, so I just copied that in. But I have a table, and all I have to do here is say, Spark read tables, and you can see that it's bound to my practice, one lake or lake house, and I can load by doing this Spark read. I'm basically using SQL to create a Spark data frame, and then I can display it. Now, the normal display would be like a table like this, which is sortable and things, but I also have the ability to create charts, and you can see over here, I can change it from like that kind of a chart to a bar chart, and it will create, convert it there. So you get a lot of capabilities here. If I use what's called a magic in here, I can change the default language, which you can see up here is PySpark, to SQL, and here I'm creating like a view to make my queries easier, and I can do selects from that view, and on and on. So I can interrupt between using SQL and Python, and anything that's created with the SQL for objects like tables and views, I can access directly from Python or even R for that matter. So now let's take a look at the data warehouse view. The data warehouse view is sort of a way to look at the data through T-SQL and get the kinds of things we're used to getting in T-SQL. We don't have, we can use, we're gonna still be writing out to Data Lakehouse format. I'll go to Data Lakehouse here, and you can see by going here, I now have a SQL query, and I use T-SQL. Top five is not Spark SQL. That's only available in T-SQL, but any kind of T-SQL commands I could use. From this mode, I can do store procedures. I can do functions. I can even see that I have security options, and refresh this, but like database roles and things, schemas, information schema is not something you would typically get in Apache Spark. So this is if you wanna have that view, a DBA certainly would be comfortable with this. SQL Server developers would be jumping and be productive right away. So you have that option. Now, I wanna make sure I talk about a couple of other things here. Data science, right? Everybody's creating models, so when you go into the data science experience, you can create ML models. Experiments are where you combine different models and you're kinda trying to figure out what's the one you wanna use. Typically, you would create like notebooks and the environment option is available there. I can import. And of course, some of these, you have this sample. Highly recommend jumping in because you can just pull any one of these examples and kinda look at how they built it, see how it works, and then you can build your own, kinda copy what they did. Finally, I wanna talk about real-time analytics because this is typically a complex thing to set up. I know, not always a fan. It can be a pain. But with Fabric, you can just easily do it, drag and drop. You create what's called event streams. So events are things that are happening, right? Could be sales, could be a click coming on, click stream, and you wanna write it typically in the background what's gonna happen. It's gonna go to event hub. But this is coming in. Think of like a fire hose. If you're Amazon, for instance, and you've got trillion rows going in, you're gonna have to get a handle on that. Event stream will give you the fire hose, but how do you query it? Well, you've got AQL, right? AQL is Kusto Query Language, and Kusto Query Language is essentially SQL or event streams, and you also have query sets. So in using those, you're able to query the stream as it's coming in, which is really powerful, and you can use that as a way to sort of analyze and see what's happening and direct that to, say, dashboards and get a handle on things. But the really cool thing is this data activator. What data activator does, they call this Reflex, is where you say, I've got this data streaming in. This could be like emergency type of stuff. This could be events going on in the world. You're doing maybe social media tracking. You wanna track if something happens like, you know, an oil tanker spills or there's some major problem because you may need to do react immediately and handle things from your business. So you can set up something here like Reflex, but I created, I pulled out an example here so we could look at it. The idea in Reflex is that I can select, like properties, what do I wanna use? This is like temperature. This is an example they give you. A medicine package is too warm, and you can see like you set the variables. If something is greater than 50, then you need to track it and you can see visuals, make it easier to see. And at the end, if something meets your criteria, you can do something like send an email, send a Teams message. And this is something that Microsoft has been adding all kinds of functionality around. So what you see now is a small piece of, I think, what you're gonna be seeing very quickly. All integration into Office is pretty much a done deal. I mean, they're working feverishly on it. And this is still just out of GA. So, you know, it's a lot of stuff coming. I've seen it in the documentation. I've been watching this for a while and they say coming soon, which is a pretty big commitment that they're gonna keep going forward with that. So key benefits of Fabric, kind of to summarize, you get a single pane of glass through the OneLake. It integrates all the different tools that you could need with the experiences to customize how you work, but be able to still collaborate. And you choose the tool that you wanna work with, the one that's right for you. OneLake integrates many different data services, and that's a crucial thing. So you can bring your Azure Synapse type of experience with your OneLake, with flat files, whatever you need to bring together. And one thing I would also say is take a look at the roadmap, Google it, go out and look, but you'll see that Microsoft is just starting with OneLake and with Fabric. Their commitment to extend this with all kinds of functionality is clear. Purview already integrates a lot of the other tools that I'm not talking about, already have integration. They've used Dynamics, that's available. All kinds of stuff is available and really just getting added to. So that's the summary of Fabric and kind of summarizing what we do at Insight. You can see from the slide, we have a lot of expertise. We cross many areas. The focus that I've been talking about is data and AI, but a lot of times that will dip into the other areas like Intelligent Edge, IoT, modern apps. You may want like a web app that works with the data and AI, whatever it is. On the right, you can see, not only do we offer consulting services, but a lot of times you may want like managed services. You may not want to have to manage your licensing or your services within Azure or any other things. We have services that will simplify that. And we have things around supply chain and other ways, even to acquire hardware and software for your on-prem solutions. That's what I've got. So I don't know, Amy, if you want to take some questions and we can go from there. Yeah, yeah. So we had a few questions. So we had a question when you were on the Microsoft Fabric slide about the ML Ops mentioned there. Is that part of the Azure ML Studio or is it separate? Good question. It is not part of ML Studio. I don't know if it will integrate or not. I don't believe at this time it does. It's actually using something called ML Flow under the covers, which is open source supported by Apache Spark, but very similar to the ML Ops that Azure ML supports. But I think this is kind of good because you have one sort of central place that is not proprietary. You can use ML Flow. Like I said, look at the roadmap because the integration, I think, with everything else is coming. Great. Thank you. Another question that we did have as well is, are the notebooks hosted on some sort of VM pool, which priced as part of the Fabric pricing? So I'm not sure if you can speak to that, but that was a question that we did get. I did, yeah. I've been looking at it. So it's really interesting. One of the goals in Fabric, and I'm glad this question came up, is to simplify the customer experience so you don't have to think about pools. You don't have to think about capacity. You don't worry about that. Instead, what you do is you kind of buy a T-shirt size and you say, I want Fabric capacity. And then that will give you so much resources, which you can tweak and adjust. But once you buy the Fabric capacity, it includes everything under the covers that you're going to be doing. So you're not going to get billed and dinged with each thing separately. It's under the Fabric capacity. Great. Thank you for answering that question, Brian. We just had another question come in about, can Fabric consider for production inference server compute platform for AI applications like LLM? I'm a little confused about the statement of the question, but you can use it for large language models. Yes, LLM. That's supported by Spark and things. So there's no reason you can't do that. You can build it, train it, put it under MLOps, and then deploy it where you like. And yeah, you should be able to do that fine. Great. Thanks, Brian. If anyone has any other questions right now before we wrap up, please put them in the chat. And if you're ready to take the next step in your journey, we have some contact information there. We would be happy to put you in contact with the right person to help you get started. I don't see any more coming in right now, but if any other questions come up, please feel free to reach out to us. But thank you. I just want to thank our speakers today for the broadcast. And if you have any other additional questions, like I said, please reach out to us or visit our website at solutions.insight.com.