View Recording: Unify Your Data with Microsoft Fabric

Suneer Mehmood 0:21 Yes, hello and welcome everyone to our webinar on unifying the data with Microsoft Fabric. So we are letting people come in. So Sam, let’s move to the next slide so that we can talk about the agenda. Sam McQuistan 0:46 Sure thing. Suneer Mehmood 0:47 All right, so this is what the plan is for the next hour, right? We’re going to talk about after our introductions each, we’re going to talk about the problem statement, right? What are the common? Challenges that organizations face for data unification needs, if at all, right? Like in a how does Fabric address that from a unification perspective, right? And we’ll talk a little bit about. The medallion architecture that has been followed in the world of analytics and why fabric? What is the advantage of having fabric? Why Microsoft is indexing so much on fabric as if. Future strategic platform for achieving A unified data analytics. We’ll talk about the competitive advantage. We’ll do some case studies right and a demonstration at a high level. On on what can be done on fabric by opening the fabric, you know platform itself, some of the work that you know we have done to show you and then next steps. So that’s at a high level. What we’re going to present here. So there is a Q&A tab over here. Feel free to type in your questions, right? And one of us will field that question and and try to answer that either. Through a wide presentation or towards the end of our presentation here, so moving on. All right, so myself, Suneer Mahmoud, I’m a data and AI architect. I’ve been with Concurrency for almost a year. I have been in the field for 17 years, I would say. So in the field of data analytics, data engineering, business intelligence. And so on and so forth. So I have seen the evolution of these architectures I’ve seen. The the evolution of these tools and platforms. So it’s very exciting to see you know where where things are headed at, you know, and especially where fabric is is one of the main. Competitors out out there in the market, you know, I would say like hands down the best. I’m joined by Sam, you know, who’s my partner in implementing a lot of this solution. Sam, would you like to go ahead and introduce yourself? Sam McQuistan 3:39 Sure thing. Thanks, Suneer. So like Suneer mentioned, my name is Sam Aquiston. I’m a data engineer here at Concurrency. Just like Suneer, I’ve been here for about a year with Concurrency. We actually started on the same day. Fun fact. But my career, I’ve been doing analytics for just about a decade now. I have been helping a lot of companies of all shapes. All sizes and lots of different industries really tackle a lot of the problems and build a lot of solutions for tons of different companies and multiple different industries. I really enjoy getting companies up to speed with where technology’s at, where it’s going, especially in today’s world of A I. There’s so many advancements that are being made, and data is the foundation for a lot of those cool things that everyone likes to talk about in boardrooms. We actually help you get that foundation set up and guide you towards that end state of having AI implemented solutions or even machine learning tailored solutions. All of these are capable or are capabilities offered in fabric and we’ll talk a little bit more about that down the slide. But that’s me, that’s Suneer. Let’s go ahead and get started on the meat and potatoes. Suneer, go ahead. Suneer Mehmood 4:54 All right. Thanks, Sam. Well said. All right, framing the problem, right? Like, so for over years, right? Like ever since the evolution of systems, ever since the evolution of like, you know, all this enterprise resource planning. You know systems that you have organizations. Organizations do need to perform multiple tasks, right? You have your operations, you have your finance. You know, depending on what kind of an industry you are, you might be a manufacturing company who is who are globally distributed. You might be a. Commercial organization who might just be operating in a few different places or might be globally distributed, but you have different functionalities and because of that you have different systems, right? Your finance systems, your inventory management systems, your HR systems, your sales systems and so on. So on and so forth. And each of those systems for the right reasons are specialized in their own category. So we wanted some way to have unified analytics on top of all the systems and over the years. You know, business intelligence engineer or data engineers. You know, earlier it used to be called as database development engineers and so on and so forth. But nowadays like everything is coming into this one common Gray area of analytics engineers, but. Although you coined different terms for this roles, you know they had the common responsibility of bringing all this data into one platform so that whoever wants to analyze the report, be it executive level or somewhere. You know, might be doing day-to-day operations, but they might be reliant on some other data, right? Like, you know, also so tactical reports as we term it. There always used to be a need to actually combine this data together, right? So. How can you actually get a data from a fragmented platform and combine them all together? The concept of analytics and business intelligence have been there for a very long while and and that has evolved. Tools have come. Database platforms have been there, you know, and and people have done done this kind of things before, right? Like bringing all your data from your SAP system or your Oracle system. I’m talking about ERP systems or your data that might be there in a. In a file based system, bring all into your analytics and then have a combined you know way of doing things. You might want to, let’s say you want, you might want to understand your sales and revenue for that matter, right? How do you make sure that like you know your data from your finance system comes? And your data from your sales or your invoicing system comes in. How do you make sure that your payment systems data comes in? So this problem statement has been there for a very long while, so we are. Over the years people have used different technologies, different architectural patterns. I would say like you know to combine all this data and have unified analytics, but here we are focusing on. Fabric at the moment and how fabric actually addresses a lot of this problems that used to be there for a very long while. So let’s move on to the next slide. All right, so. You know, taking, taking one step ahead from the from the problem statement that we discussed about, right? Like in order to achieve what I was talking about, people had to have different services or different, you know, services was going, you know, from. Ever since the formation of Cloud infrastructure, right? But before that they used to be like your your applications or your tools or your your servers that used to achieve all those functionalities. You need to have a ETL application. Which would extract the data from your source systems. You need to have a database application which stores the data. You need to have a reporting functionality which kind of analyzes this data, right? And these are at the 100,000 foot level, those people who have been in the analytics. Still, I mean analytics field can easily understand this is at a very 100,000 foot level, but in order to achieve all this. People had to use different different applications or services as you call it right and and over the period you can you might have heard of you know SQL Server Integration Services you know which is primarily Azure or or Microsoft and ever since Azure Cloud there has been this. Azure Data Factory Synapse Analytics, Synapse Data Pipelines which would actually pipeline the data into some sort of analytics layer and then from there there has been this concept of like querying your data on a database that is like. Performant enough so that you can get all of your historical data aggregated in some form. So Long story short, people had to use different services to actually get this all like you had to get a database stood up. You had to get a ETL application stood up. And then you had to have like a Power BI service there. So you had to spend a different four different services. You had to have licenses for all those different services. Also you had to worry about how these services actually interconnect and work seamlessly. Also, in the past 10 years, the the concept of big data has evolved. Data has moved from a structured way of representation of data earlier, like all the systems that store the data knew what sort of structure it has. Now we’re talking about Internet of Things, right? Just think about your let’s say Google Nest or some sort of smart device that you have at home, right? Like the kind of data that stands is like kind of in a semi structured fashion, right? Like your click. Stream data. It’s a semi structured data data that you get. You cannot kind of you know prescribe the data has to be in certain structures. So gone are the days when you used to only get data in a structured fashion. Now you’re getting data in a semi structured fashion as well. So then we started the concept of data lakes, right, so that you could store all of this. Data as files and then we are talking about the evolution of data into unstructured data. Think about videos and and images that needs to be analyzed, right? Those are the unstructured data. So we had to actually form something called object based storage or file based storage as you call it, right? So. So it is database on one end, it is like data lake on the other end. Then you need to have all this ETL methodologies that would actually bring source the data from one end, you know, do whatever munching that is required, drop the data. So you you need to have specialized services for all of this. Fabric is bringing all of this together in one ecosystem so that you do not have to actually spend four different licenses. You do not have to worry about, you know, integrating these services and making sure that they work seamlessly, which used to be a big tab. For data engineers and so on and so forth. So plus on top of that there is copilot in fabric. We are all getting into the realm of AI now like hey, how can I as a citizen analyst I can go, I want to go in and you know what I want to just. Query my data in my natural language so that I can get the response that I needed. I don’t want to depend, you know, rely on an analytics team. Rather like you know I can ask my questions in a natural language and I want to get that data. So that is there in Fabric too and and I you know Fabric brings in a very interesting concept called One Lake which I already briefly touched upon like in what is a data lake to store all of your data. But the concept of One Lake is that like you know, Fabric lets you connect to a lot of different sources. Not Microsoft, even outside of Microsoft as well through this concept called One Lake, which we’ll talk more about and so that you’d not have to even move the data from there. Let’s say you have data in, let’s say you have a multi cloud environment. You have Azure, you have AWS, you have Google Cloud. So Fabric has a concept of one lake where you wouldn’t even have to move the data from there. You can actually connect to that data sources through the concept of one lake and still do all of your compute and analytics in one layer. So that’s what one lake is about. Purview, which is like your governance, like you want to understand where is my data coming from. If I change this particular field, what will happen to my report? Will my report break or not? Or or like you know, hey, I’m not finding these numbers right like and I don’t know if I can rely on those numbers. I want to really see where is my data actually coming from so. Those kind of lineage impact analysis and also classification of your data like you might have sensitive data in your system, you might have PII, BCI data like in a personally identifiable information in your system. I want to make sure that this data is protected, right? How do I make sure that all this governance? How do I make sure that what all do I have in my ecosystem? First of all, I want something to go and scan in all my metadata, all my information that is stored in there and give me information. I have sales data here, I have finance data here, I have customer data there. That could be a I want to make sure that there is no credit card information, there is no personal information. I want to classify all of that and then lock it down so that only the required users have access. All those concepts are brought in together in a unified fashion within. Within fabric earlier, we used to have different tools for that moving on. All right, so this is like a concept of like, you know, over the period as I as I explained, right like. We used to have different patterns of moving the data for different reasons. Let’s look at the green circles over there. I guess it is green. OK, yeah, we have Spark, we have T SQL, we have KQL. We have analytics services, so Spark, whoever has been in this analytics side would know Spark is a big data processing architecture as I mentioned, we have started getting. This semi structured data, unstructured data. How do you process all this big data in a distributed fashion so that it’s very efficient? You can process a lot of this data in parallel. Think about your V8 engine or V6 engine. You have a lot of pistons going on to power your machine, right? How do you? Parallel process all this data. So that’s what Spark is about and over the period there has been competitors in the market like Databricks and several other vendors have been there like Synapse Analytics within Azure itself has been. I’ve been using Spark, so you have Spark within Microsoft Fabric to move the data in a big data parallel processing fashion. You have T SQL or Transact SQL where that’s analysts like bread and butter, right? Hey, I have Power BI, so how does Power BI? Extract data from a relational database which has been there for a very long while. Like your MySQL, your SQL servers, your Oracles, your DB2, your Neteaser, all of those things. You want to analyze this in a transact SQL format. So Fabric has T SQL component within that and the way in which you store the data is. There is a service called Data Warehousing there. Just to touch upon like in a Spark, there is a service for Data Factory which is like your GUI or Graphical User Interface based pipeline where you can build this pipeline. You can actually write your Pi Spark, your Scala notebooks and so on and so forth. Different languages by. By the way, to do all of that, you have your serverless compute, you have your custom query language, KQL, right? In order to actually get real time intelligence of your data, let’s say you have streaming data that you want to get to you’ve been using. KQL has been around for a very long life. And you have KQL component also within Fabric. You have your analytic services like Power BI. Power BI has been the top, I would say the top notch visualization tool for four years now. So Power BI is there in Fabric. You don’t need to get separate Power BI license if you have Fabric. And as I was mentioning, if you have data not just in Azure, you have like Amazon, you have your GCP, Google based data. You can actually connect to all of this data using one lake. Let’s say you’re sourcing the data from Azure SQL DB or any other sources you can actually have. Data Factory bringing this data in. You can also mirror your data, not just Azure SQL DB. You can actually create synapse links to actually mirror your data from other databases other than SQL DB as well, like maybe Dataverse or Cosmos databases and so on. And so forth. So you have all the services under one ecosystem which lets you have a holistic way of handling all that seamless integration, not having to spend licenses for separate services and it’s being spread scattered around, if that makes sense. Let’s move on. All right. Medallion architecture. It’s an interesting top topic. Sam always says like you can talk about it for 1515 minutes if I, you know, that’s a favorite topic like you know that I have because I have seen the evolution of this architecture over years. So. The term medallion was coined by Databricks in 2010s, early 2010s, but the concept of this has always been there, right? Let’s say you are. Bring data to your analytics system. You don’t just bring it to one analytics layer, right? Like you want data to actually get promoted from one layer to another forward, right? Like. So let’s let’s take a hypothetical scenario of your your. As the picture shows here, there is a CRM, enterprise resource planning, there are Cloud databases, there are data lakes where you could have structured data and semi structured data where there is no guarantee that. You know this is a structure of the data that we that is being stored and real-time data streams absolutely like in a again a semi structured data you can you don’t have no control of like you know hey the structure of data that is coming in. So back in the days when we used to extract data from these sources. We used to have the concept of ETL or Extract, Transform and Load, right? So we would extract the data from the sources, do the necessary transformation like it could be like maybe calculating some columns or maybe like cleaning up some data, maybe addressing the nulls and. And maybe deduplicating that data or maybe marrying the data with some of the data to make it more meaningful and then load the data. We used to follow this pattern. So the problem with that pattern has been as the data grew, the needs for analytics grew. It used to have a footprint on the source system. All these ERPs have a. If specific purpose right? Like you know that CRM has a purpose, ERP systems have a purpose like your SAP or Oracle ERP system. So the footprint of analytic system is so much on this. Meanwhile the transformations are happening. You know, that kind of impacts, you know, the primary purpose of that ERP system, right? Like, you know, hey, I’m supposed to register my sales, I am supposed to register my invoices or, you know, write my transaction data. I cannot. You know, reserve my data’s I/O for analytics, right? So that’s the that’s when like you know, we move from the ETL concept to a concept called ELT concept, extract, load and then worry about your transformation later, right? So that you have minimal footprint. So if you’re extracting, loading and then you’re doing a transformation means like you know what? You’re not worried about what data is coming in, you just want a area to just drop that in. That’s what bronze is, by the way, right? So and and as the medallion name implies like bronze being like you know you’re you’re maybe you know not so great layer, silver may be a level higher, gold being your pristine layer. So that’s what the concept is here. So what happens in a silver layer is like. Data can get filtered, data can get cleaned, augmented, standardized. Because let’s say for example you might have your customer data that is coming from your your your CRM system or you might have your customer data that is coming from. Your your sales system, right? Like you know how do you actually standardize so that you have a customer master data over here, right? Like you know. So those are the use cases where you would like to actually deduplicate, clean, augment all of this data. That’s what the silver layer is for. But is a silver layer really a layer where you could do fast analysis? No. Well, technically you can, but architecturally not so recommended because gold is your layer where you can actually have the most. Performant solutions and also the data stored in such a way that your faster analytics is always given priority. Let’s say for example you might not just store the data in a data lake with a compute engine there, you might just store the. Data in a warehouse where all of your data is indexed, compressed and your data could be pre aggregated so that the report can perform. And and so on and so forth. And also if you have heard about the dimensional model or a star schema model as you would call it, you have your centered fact table which has all the KPI’s that you want to analyze for a specific context and all the dimensions surrounding it which could be. Those elements that you would slice and dice the data that you would like, you know, analyze those KPIs with, right? So for example your customers or your products or your geography locations or your like you know, date dimensions and and and your product categories and so on and so forth, so. Those are arranged in a dimensional model or, uh, you know, uh. Can be done in a in in that layer, in a gold layer. Typically silver is more like your cleansed layer and gold layer is more like your pristine layer. So think about raw, you know, bronze being your raw ingredients for food, right? Like silver being like you have your prepared food, but gold is like. Your food being presented on a dinner table, like to consume in a very performant fashion. BI and reporting are mostly done on the gold layer, whereas machine learning. And AI can spread around because machine learning and AI, they don’t just rely on the pristine report level aggregated performant data. They might want to get into a little bit of like you know. Unprepared data as well in order to do some mining in order to find some patterns, right? So based on the needs, machine learning can go into the silver layer and even bronze layer. I’ve seen machine learning data scientists looking at the bronze layer to understand, hey, how’s the data coming from all the systems because you have the raw. Representation of that data in the bronze layer. So yeah, that’s what at a very high level medallion architecture is. I will hand it over to Sam to take it from here to talk a little bit more about fabric and do a little bit of demonstration. On what we have done. So over to you, Sam. Sam McQuistan 27:04 Thank you, Suneer. So you might be asking yourself, after all of this explanation, you know, what makes Fabric unique? Why should I choose Fabric as my analytics platform of choice? Well, I think that there’s a lot of reasons, and Forrester Consulting has already done an economic analysis and impacts for businesses and organizations alike. That have adopted Fabric up to this point, and understandably so. We’re in the early days of Fabric adoption. However, Fabric, Microsoft recently announced, was one of their fastest growing platforms ever. And for analytics platforms, it is their fastest growing analytics platform, even faster adoption than Power BI initially. Nearly a decade ago. But for Fabric, I think the big key selling points are that it’s a it’s a unified platform at the end of the day, which is what we’re here for. You know, we’re talking about how Fabric can unify your data sources and unify your analytics, but it also unifies engineering governance. And analytics all in one place, which traditionally were all separate services or all separate platforms. And like Suneer had mentioned, even just 5-6, seven years ago, we were talking about having lots of different vendors for different responsibilities within the analytics stack and that though that can lead. To bills stacking up over time, it can lead to reduced ability to monitor. So having that unified platform for everything underneath one subscription in Azure, especially for existing Microsoft customers, is generally a huge win for them. When it comes to business outcomes, one of the biggest ones is that you have faster time to. Insights. You have a lot of streamline costs within an Azure line item and you’ve also improved analytics collaboration. Getting faster insights, especially in today’s economy, is very crucial for being able to know that you’ve made the right decision or that an executive is proposing the right solution to. To a problem that you’re facing in the business and when it comes to analytics collaboration from the engineering standpoint, being able to see what items your co-workers are working on, what changes are being made, having a lot of version control tools built into the platform is huge for being able to trust. That the data has been transformed in the right way, or that the data is being treated correctly. It also Fabric also solves a lot of common pain points, you know, especially with vendor lock in or vendor spread. You know, Fabric doesn’t really encounter the same problems. A lot of businesses and organizations have had issues with having a lot of tools or having really slow reports or dashboards depending on the way that the previous architecture had been set up and observability. Like I mentioned, when you have that all under one roof, identifying a failure throughout the entire ETL or ELP. Pipeline for data transformation is huge. You can see whether a bronze piece failed. You can see whether some custom transformations failed, whether there’s some other technical transient issues that are occurring. All of those can be identified within the same queue as opposed to having to 1st identify where in the process. Something broke and then understanding what system it broke in and then being able to diagnose some of the semantics. Everything is now under one roof and it gives insights and abilities to everyone in the analyst and engineering groups. It also increases productivity. Forrester found that analysts and engineers alike saw huge efficiency gains. When after the implementation was done, they were it was a lot easier to find data, it was a lot easier to query that data, and it was a lot easier to create a report on top of that data with a lot of functionalities that traditionally you would have to have an engineering specialist come in and optimize, especially when it comes to. How reports are built on top of semantic layers or business model data. You don’t have that problem with Fabric when you use a what’s called a direct lake connection, which we’ll get into in the next slide as that’s another huge competitive advantage for Fabric and. Similar to the Azure Build consolidation, a lot of organizations see that they reduce their infrastructure costs and it also reduces the time to set up that infrastructure because Fabric is built for mostly tailored towards a citizen developer or citizen data engineer. So when you do have reduced infrastructure costs and all of these are seen, it’s really easy to adopt Fabric and see what the business possibilities are when it comes to return on investment. Forrester found that the average ROI for Fabric implementations when accounting for scaling down existing platforms and migrating equivalent jobs. Over in the fabric, you can see your return on investment be achieved in just as little as six months. In the next slide, I want to talk a little bit about some of the competitive advantages. So what makes Fabric unique? You know, we talk about how Fabric is this great platform and how it has lots of features that are beneficial for business leaders, engineers, analysts. But what makes Fabric unique and stand out from the rest of the competition in this space? Because there’s tons of. Options First First off, it’s a unified SaaS platform. Everything is under one roof and it takes very little time to set up the initial architecture and infrastructure. Once you set up your Azure resource for Fabric, all you have to do is grant access to your data architect, data engineer, whoever’s going to be responsible for the platform. And they can go in and start to build out the architecture itself. And since Fabric operates under a what’s called SKU licensing format, you you can rely, you can, you can rest easy knowing that you’re not going to exceed a lot of. High high level costs associated with spinning up a lot of compute clusters. There’s always going to be a Max limit based on a budget, so it’s really easy to set a budget for your business or for your analytics department, operate within that budget, and then optimize from there. And like I mentioned, this platform is geared towards the citizen developer, but that’s not to say it doesn’t have a lot of technical capabilities that other platforms have generally exceeded in. Next, I want to talk about One Lake. This is a central data hub for everything on Fabric. One Lake has been built for. Build as a place where you can spin up nearly an infinite number of storage accounts and different storage accounts at that. Fabric now supports Lake House configurations, which are very similar to Azure Data Lake containers. They also support warehouse configurations which are. Going to be treated more in the traditional RDBMS sense, so it’s going to mimic roughly what a database would excel at, which is things like indexing or running stored procedures. Or for the casual analyst who knows SQL and can write their own SQL queries, they then have the option to. Everything in SQL on top of a dedicated analytics layer. All of this is stored on the same basically Azure Storage account, and underneath that is where you’ll find each of those individual resources. Another new feature with Fabric is that they now support SQL databases in public preview. So. This is not a generally available feature, but you can see that Microsoft is trying to integrate a lot of different storage techniques and a lot of different storage technologies to allow you to have more of a wide breadth of ability on your data team as you compile them. So you can hire analysts that are really good at SQL. You can hire. Analysts that have maybe dived into a little bit of Python or understand how to write like a Jupiter notebook. And you can also hire someone who maybe has more business like experience and understands what the semantics of the data model are, but maybe wants to learn. I think Fabric is a really good play. Ground for a lot of that, and that’s why One Lake makes all of this extremely easy. You have a wide range of options available to you as far as implementation goes, and it’s all stored under one roof. Next, developer collaboration is really good in Fabric when it comes to working with fellow analysts and engineers. Like I mentioned before, you can actually see what everyone is working on. So if you’re in the same notebook, there’s locking conflicts that step in to make sure that one edit doesn’t overwrite the other and that you’re maintaining a master. Version of all of this code. There’s also built-in CICD options and version control as well that we’ve been experimenting with with our client that just went live in production. And that’s not to say that you don’t have other options available to you. I think a lot of advanced developers would prefer using options like Git or using an Azure DevOps repo. Those are both options available to you as an organization that’s adopted Fabric. So you’re not just limited to the tools that are built into Fabric. You also have external connectors that you can use for some of these really important functionalities for develop for developing good developer culture. And then finally, integrated AI and ML is really nice for Fabric because you can make use of some existing workloads. If you’ve been diving into machine learning in the past and you have different vendors that are set up to take advantage of data sets or data structures, all of that is available in Fabric. Even if you’re not familiar with it, it’s really easy to get in there and start playing around with it. There’s no extra cost associated with using these features, so you can see how they run and assess whether or not you want to bring those over into Fabric. But it also does play well with external providers because those data sets in Fabric are still available like your SQL databases. Databases or Your Data lakes have passed. As long as you’ve set up the proper authentication modes, you can use these with all of your external vendor tools as is, and use Fabric as an analytics platform. There’s truly a wide range of features available within Fabric, and Microsoft continues to promote new features. Into GA every week, every month. So if you haven’t assessed Fabric within the last even 6 to 12 months, things may have changed and Fabric may be a good option now just given all of these different points. So I’ve been talking a lot about the client that we recently got into a production environment and deployed. And I think that this is a really good case study to understand how Fabric is built for truly organizations of all shapes, all sizes, all industries. There’s no limit to what you can build out within Fabric, but what we were able to do was. We we got contacted by a client who is a very large manufacturing company and they have several subsidiaries. They’re a global business, so they have ERPS databases and subject matter experts in different time zones all the way around the globe and mostly this is this company. Oper. On multiple different ERP systems. So we have some that are very well recognized names in the space like Oracle and SAP and we even have some ERPS that aren’t as well known but are more regionally focused or are for or are designed for more niche manufacturing markets. So we were dealing with a lot of different. Different vendors and a lot of different systems. And in a traditional analytics stack, without Fabric having all these connectors under one roof or built in, it may require you to set up multiple different vendor subscriptions for ELT tools or storage accounts or even have a separate Power BI licensing structure. Let’s say you wanted to use Tableau as an example. You know you can see how the amount of subscriptions that you would normally need for these kinds of projects can be reduced with the functionality that Fabric has. But some of their common pain points are things that a lot of organizations see and being able to being able to open the eyes to what. The possibilities are, I think can show you why Fabric is a good platform. This client had really slow times to reporting if they had an analysis request that they wanted to fulfill or there was a question that they were trying to answer as it pertained to their business. The traditional flow would be that requester would then go to IT and then IT wouldn’t figure out where some of those responsibilities lie within the individual subsidiaries. And if it’s a multi subsidiary effort, then a whole committee gets formed to solve these kinds of big ticket items that executives are wanting to react. Back to on a much quicker basis. So once you start to involve all of these different personas and all of these different subsidiaries, the logistical hurdles can really start to surface when you don’t have a unified analytics platform. You might be looking in different ERPS, you might be doing data exports from databases. Or even a combination of the two or bits and pieces of those options. And when you have those different systems or those different rules, it’s really difficult to get to that master data set that a lot of businesses are looking for when they do a unified analytics platform. In analytics, we call this data conformation, which means we take data sets that might not look the same, might not behave the same, but we model them in a way, and in this case in fabric, so that these data sets can merge together, represent common business terminologies or common business measurements and be deployed. Widely to anyone in these subsidiaries or anyone within the holding company. They also lacked a lot of ability to analyze proprietary data and make quick decisions and with operating silos in IT, sales, finance, marketing, you know, you name it. It’s an executive decision that has to be made that they need trusted analytics and they need that be able to have it fast, even on a daily or real-time basis. Now we take all of those factors, put them together and think about how 2025 has been such a turbulent market, especially in manufacturing. There’s a lot of. There’s a lot of things going on as far as pricing goes, a lot of issues with getting parts supplied and that really made it that much more effective for them. Now that this has been fully implemented, these decisions can made can be made a lot quicker than they did before, sometimes even reducing the. Time to get data sets compiled from weeks or months or even upwards of 1/4 or a years long project to in as little as a couple of days or even a week’s worth of time having an analyst or a data team tasked with solving those problems. So you can see how when you have a large company or you have a lot of different subsidiaries operating just slightly different from each other, having that unified analytics platform can really help. And that’s where Suneer and I would come in and say let’s, let’s look at the broader landscape of your business. Let’s understand what systems are out there. We would assess. If. Is the right option and if not, propose alternatives. And when we do decide that Fabric is the right option, we start working with you to identify who in your business has the knowledge to understand some of these systems. If we don’t already have that, we can also work with you to understand common data scenarios where sometimes data doesn’t behave the way that we expect it. We’re here to help organizations of all shapes, sizes and industries be better with their data and make more informed decisions. Because at the end of the day, if you don’t have the data available to you to make a decision, or you maybe have data that is. Incorrect, and you make the wrong decision. You can see how that opens up a wide range of possible outcomes, both good and bad. So in the next steps, I’m going to actually show you a little bit of the Fabric portal and show you what it’s like to work within this platform on a daily basis. Before we do that, I want to show you how all the things that we talked about as far as architecture and data format play into a Fabric implementation and one of my favorite features in Fabric which allows me to visualize. And explain to business stakeholders or analysts or engineers how the architecture was set up. With a medallion architecture, you generally will have a bronze, silver and a gold step, each containing different operations. But all of this data ultimately gets sourced from different systems and put into similar categories. So as you can see here. Here we have an we have a ETL orchestration engine that kicks off all of these pipelines on a system by system basis. So each of these jobs is segregated from each other. In the bronze section we have all of these different unique ETL jobs to handle data from source and get it modeled into the modeled zone. And then the silver transforms are going to take that data from each of these systems and treat it as the system would normally intend it. Or we try to conform this data into the shape that we need it to be in at the reporting layer. And then in the gold we do a simple lift and shift to get that data from the silver modeled zone with update. Dated or new data and put it into the data set or or in a warehouse in this case to take advantage of things like indexing or data store or efficient data storage storage and then build a Power BI report and a semantic layer on top of that for data to be analyzed in dashboards reports. Various analytical scenarios that you might have, and we’ve also taken advantage of a couple of different features in Fabric like database mirroring. So we have a solution where we take data from all of these different sources and there’s some elements that need to be manually configured or assigned to some of. These things like customer data or customer site data like where a product is being shipped from or shipped to or even billed to. A lot of this common data gets written into an Azure SQL database which could also technically be done on Fabric, but we also have mounted a Powerapps application. On top of all of this master data so that business users and subject matter experts for this particular organization can assign custom values that they may want to build reports on top of. So think about if you have a whole list of SKUs that a particular organization might sell. You might want to assign things like a product hierarchy to it, different families of products that have been sold, and then do analytics on top of those individual groupings. We’ve built forms that allow you to assign different stages of a product hierarchy to an individual SKU in an external or an externally. Racing application and that all automatically on a live basis feeds into Fabric so that that data can then be used in the reporting layer or in any layers past the silver layer. And then on the left side here, this is an example of how we’ve actually simplified the folder structure in Fabric. All of these items can get grouped together in a common format. It’s easy to find items. It’s easy to identify where some of these items live, and then you also will have a dev and a test, a prod workspace. You may even have other workspaces that are more geared towards a specific. Department within a business. So all of these are different options to store your data and no one-size-fits-all, but it’s important to know that you have options available to you and then there are a ton of different unique ways that you can build your Fabric environment. So I’m actually going to give you an idea of what the Fabric workspace. Looks like here. So when you sign in to Fabric for the first time after your SKU’s been set up, you’re going to have options over here on the left side to navigate through workspaces, view all the different storage accounts that are available to you. The monitor hub is where you can actually see the unified. Log view, see what’s running, what’s failing, what’s succeeding, and then dive further into those errors. If you do see failures, we have a real time data hub where you can see live data coming in from each of these different source systems. And then understand what’s being inserted. And then we also have workloads. But for the time being, we’re going to go in and talk about some of the different features that are available to you and show you how easy it is to actually set up an architecture for yourself in Fabric. So when you come into your development workspace and there’s nothing in here, you have the option. To either start from scratch on a visualizer, or Microsoft has included a lot of these different predesigned task flows, and these task flows will allow you to assign or tag individual items like notebooks or SQL scripts or analysis, reports, dashboards, storage accounts, everything under the fabric. Work. Space can then be assigned to its appropriate location within, say, the Medallion architecture. But the Medallion architecture is not all you have. You have things like basic data analytics where you’re just getting the data, storing the data, visualizing it, and then maybe doing some tracking on top of that. You also have the option to do event analytics where you’re. Be getting data from a live data source and you’re also getting it from a local SQL database. But you need to indicate that those are fundamentally different things, and you can assign those to different steps and treat them independently from each other. There’s also options for basic machine learning models, event medallion architectures, translitical architecture. But for our use case, I think for a lot of traditional analytics implementations, the Medallion architecture is what we recommend. And Microsoft gives you kind of a guide here for how you want to design or work through all of these different responsibilities in the analytics workflow. And assign them to each of these tasks. O when you’ve decided on what task flow you want to use, if you want to use one, you simply hit select. The task flow gets generated here in the user interface and then you can start to actually build out some of your folders and add your items to those. So let’s say we wanted to add a new item and we wanted to start at bronze. You have a ton of different options to actually get data out of your source systems and into Fabric itself. So you can do a traditional copy job, which if you’re familiar with Azure Data Factory, it’s going to be the exact same engine. It’s going to be a lot of the same configurations. That are available to you, but now it’s included in Fabric itself. You also have options for Dataflow Gen. one and Gen. 2, which are what Power BI is traditionally used for getting and transforming data. There’s also event streams which will connect to a live data set and incrementally ingest all the new events that come in. There’s also options for mirroring your databases. So if you have an Azure SQL database or soon, there will be options to actually mirror from Oracle databases as well, and I think SAP is potentially even something that is in a preview phase. You can get data changes live from your data set or your database mirrored in the fabric in near real time. And then there’s also some options that I wanted to talk about with like Databricks and I believe Snowflake is. Yeah, Snowflake is also in Georgia. So you can also let’s say you’re. If you’re not new to the analytics world, or you have platforms on Databricks or Snowflake, but you’re potentially wanting to explore migrating into Fabric for Power BI purposes, you can actually mirror your entire Databricks catalog or mirror your entire Snowflake catalog with just a few different clicks of a configuration and from here. You can. Build everything that you want in Fabric, but also maintain some of the code that you’ve had in your in your previous analytics stack, whether that’s Databricks or Snowflake. So there’s tons of different options for getting data out of all of these source systems, and when you create an item, you can then tag it into the bronze step or the silver step or the gold step. There’s also options for. There’s also different options for storing this data too. Like I mentioned, we have data marts that are a legacy carryover from Power BI. We have event houses which are automatically structured to better handle live data or or really big live event data that’s coming in. We have lake houses which are going to support structured data, unstructured data, pretty much data of all types, even supports drag and drop for say like Excel files. If you have a lot of this master data living there today, you can write, you can then write things like notebooks on top of these Excels Excel sheets and then get that data into your data environment as well. You also have warehouses that are going to more efficiently handle structured or model data in your particular environment. Think of things like indexing or running stored procedures or doing merge operations on highly structured data. All of those are things that are available in the warehouse configured. And then there’s also in preview a SQL database option where you can stand up your own SQL database underneath the one lake container for your environment and do all of the different kinds of SQL database development that you would normally do. On a on like an RDBMS. So there’s tons of different options for storing data. There’s tons of different options for getting data out of these systems. And then when we prepare data, there’s even different options for for manipulating that data. As I mentioned, you have all the options available to you in a traditional RDBMS sense, but you also. Have the options to schedule airflow jobs if you are familiar with those. You can do custom Azure Data Factory jobs if you so please. You can even do notebook activities which are going to roughly mimic what a Jupyter notebook does. Or if you’re familiar with Databricks, I know Databricks is very popular in the notebook. Transformation space. You have almost all the same functionalities as that platform here in Fabric, and it’s going to be nearly identical in terms of use cases and being able to use these features. So there’s a lot of different options that are available to you in Fabric. That normally would take a lot of different vendor accounts to accomplish. So This is why we feel Fabric is a great option for you to unify your data. It’s easy. It’s geared towards a citizen developer. You have lots of different technical options available to you. And when you’re building out a data team, it allows you to think about hiring employees of all different backgrounds and experiences. You’re not looking for the typical cookie cutter candidate. You can build your team how you please, whether that’s focusing on the knowledge, focusing on the technical implementation, or focusing on. And getting all your data into one place. Fabric has all these features that make it extremely easy for you to do. And with that, let’s talk about what’s next, right? So we’ve talked about all of these different capabilities of Fabric and what you can do with it. But ultimately, Fabric is going to help you give yourself that strategic edge over some of your competitors. When we talk about what you should be doing versus what your competitors might be doing, or maybe even what you’re doing today and you want to change, don’t be the competitor that is sitting and reconciling data sheet data spreadsheets. All day for weeks or months just to get one single answer. You could be forecasting your future sales, assuming that you’ve set up the architecture first that can be trusted in Fabric. And if you’re wanting to migrate from another solution, like I had mentioned earlier, Fabric runs alongside Azure. Without this rip and replace mindset, you can. Active. Continue to use what you’ve had before while standing up the features in Fabric that will allow you to easily migrate over. Or if you’re on a platform like Databricks or Snowflake, you can actually bring everything in from that platform over into Fabric or even rewrite it into Fabric in a very similar manner as you’ve been doing in those platforms today. Fabric is also in active development. There are new features and connectors being released frequently and like I mentioned earlier, if you haven’t maybe evaluated whether Fabric is the right option for you over the last 6 or 12 months, it may fully be compatible with your environment giving given your system mix or the different technologies. That you utilize even in the last six months, there’s there’s been improvements where we would have wished we had it even a month earlier because that could have changed the whole game. And then finally you get AI out-of-the-box. So copilot is automatically built into every single fabric SKU that you can purchase. Just through Microsoft today. So whether you’re someone who wants to use it for engineering purposes and you’re using it as a guide to set up architecture standards in engineering, or you’re an analyst or an executive and you want to ask a question in English or English language or your language of choice. What a particular data point might look like. You have those options available to you in any licensing structure, whether it’s the F2 base license or all the way up to F2048. You have options for copilot in every single fabric SKU built in with your subscription. And with that, I think we’ve mostly finished this. There’s a couple of additional points that I think Amy, if you have, if you have some ability to talk about the complimentary next steps that we have available to you, that would be awesome. Amy Cousland 57:19 Yes, hi everyone. So I dropped the survey link into the chat there. We have two different free assessment options for you. One being the data, sorry, the data readiness and fabric fit assessment. With some notes on what that is and also a fabric value discovery session. So those are 30 minutes complimentary with our team to discuss your initiatives around data and fabric. So even if you whether you’re interested or not, we definitely like your feedback from today’s event. So if you can please take a second to fill out that form. uh We really appreciate it. Otherwise, if there’s any questions, if you want to chat them, otherwise we’ll be wrapping this up. Sam McQuistan 58:08 All right. Well, with that, thank you everyone for joining. It’s been great to tell you all about what Fabric is and what it can help you do. If you want us to stick around and talk with you about some of your use cases, we’re happy to do so. Otherwise, we would love to have you guys. Contact us to talk for even 30 minutes about whether or not fabrics right for you and what we might be able to help you with. Amy Cousland 58:31 Thank you. Suneer Mehmood 58:32 Thank you everyone.