Fabric is Nothing, But it’s also Everything

Rob Collie (00:00): Hello, friends. Today's episode is the first ever Jam Session, that's our working title for it anyway, where Justin and I pick a topic and we talk about it. Whereas, Christopher Walken would say, "We really explore the space." For the first time out, what better space to explore than that of Microsoft Fabric?

(00:18): One of the reasons why we're particularly excited about this Jam Session format is that, on the one hand, you've got me, the historian, sociologist type of perspective, and on the other hand, you've got Justin, who actually knows things. Let's be clear, there aren't many Justin Mannhardts walking around in this world.

(00:34): He's one of those people that seems to have more hours in the day than other people and he uses those very efficiently. He's absorbing and digesting a tremendous amount of information at all times, and he can turn around and replay it to people. He gives it a human voice. Those are the kind of hybrids we need. Certainly, the hybrid I need. You can keep your ChatGPT. I've got Justin. Ask him anything.

(00:56): Unlike ChatGPT, if he doesn't know the answer or if the answer is not yet knowable, he'll admit it. ChatGPT goes right ahead and makes something up. Silly humans can't tell the difference, can we? For example, I can say things like, "Well, it really seems to me, reading the tea leaves, that Fabric is like this and it's headed in a direction like that." And then, I look at Justin and go, "Is that right?"

(01:16): And to give myself fair credit, I've got a pretty good batting average on things like this. That dynamic is already very much on display in this first time out. We talked about things like what we think its impact is going to be for small to mid-market organizations. Talked about governance, Purview, OneSecurity. Talked about how Fabric really is nothing and everything.

(01:36): I think we talked a little bit about what good news this is for the Power BI crowd. I'm positive we didn't talk about it enough, so we're going to be circling back to that shortly in a future episode. I also call out a couple of times during the conversation that these questions that we're answering, some of them came from the community.

(01:51): So if you have a topic or a question you'd like us to address, please do share. Hit us up on social media, or even better, search for the Raw Data By P3 Adaptive LinkedIn group. If you search for Raw Data By P3 Adaptive, you'll find it. One of the happiest things I've been doing lately is approving membership requests for that group. So much fun. All right. I think that's enough yammering. Let's get into it.

Speaker 2 (02:15): Ladies and gentlemen. May I have your attention, please?

Speaker 3 (02:19): This is the Raw Data By P3 Adaptive podcast with your host, Rob Collie, and your co-host, Justin Mannhardt. Find out what the experts at P3 Adaptive can do for your business. Just go to p3adaptive.com. Raw Data By P3 Adaptive is data with the human element.

Rob Collie (02:44): All right. Well, welcome to the first ever Jam Session or Jaw Session format of the Raw Data podcast. Today, we've decided what better way to start than to talk about Fabric. Fabric is the new hotness.

Justin Mannhardt (02:57): Whatever do you mean?

Rob Collie (02:59): Exactly. Whatever do you mean is why Fabric is the title. And that should be it. The way I see it is, in 2010, Microsoft decided to what we would now call, "Citizen developer-ify", democratize the OLAP analytical database space. It was the apex predator of IT technology.

Justin Mannhardt (03:22): Right.

Rob Collie (03:24): Just the absolute hardest freaking thing to do was MDX models from Microsoft. And yet, when you got them working, they were the best thing ever. The one or two times in history that they actually got working.

Justin Mannhardt (03:37): One or two.

Rob Collie (03:39): You had to adjust your standards for working in order to call things a success. A Manhattan Project-level long shot that this would work, and it did. Not only did they bring that down to level of people like me ... I'm no dummy, but I certainly wasn't able to MDX it back in the day. I tried many times.

(03:59): They brought what was arguably the hardest thing in IT technology down to the level of the data-geeners. The VLOOKUP and Pivot crowd like me. It was just a miracle. What we're seeing from Microsoft is that they're like, "Well, why can't we do that for everything?" If you look at their track record, it's been pretty good. They did it for ETL.

(04:21): Power Query is another just unbelievable miracle. These are things that you wouldn't necessarily have bet were possible, and they've done a great job weaponizing that with flows and automate and blah, blah, blah. There are other components of the Power platform like Power Apps. I've never written a Power app, so they didn't quite bring it down to my level, but this is their strategy now.

Justin Mannhardt (04:45): It's their mission statement.

Rob Collie (04:46): They're going to do this for everything and Fabric is just a massive expansion in this ambition. They're like, "Let's take it to everything." That's even data engineering, what we used to think of as data warehousing, all the stuff that you have to do upstream from AI.

(05:05): Why not go declare war on everything that's left and bring it down to this really, really, really powerful audience? We've called them many things, the data-geeners, the VLOOKUP and Pivot crowd, the ambassadors between business and IT. It's very exciting, but as usual, it's also really confusing.

Justin Mannhardt (05:28): Ding, ding.

Rob Collie (05:31): For example, Fabric, as far as I can tell, isn't anything.

Justin Mannhardt (05:35): That's right.

Rob Collie (05:36): There's no Fabric feature. You don't sit down and do some Fabric. It's a marketing umbrella term that even now encompasses Power BI. Confusingly, when you go to Power BI and you load a report, you get a brief flash of the Fabric icon. There's still Power BI icons in various places. It's like, "Okay. Which one is it?"

(05:54): It's like Fabric needed to stake its claim over Power BI. I think it's smart that the Fabric icon shows up. It's an umbrella statement for all of this democratization that they're going after. Do you still agree that that's the narrative here?

Justin Mannhardt (06:09): I was thinking about when you said, "As far as I can tell, Fabric isn't a thing." I just want to lean into the Microsoft naming conventions struggle for a minute. We joke internally about data fabric, data mesh, data cashmere. A sort of party line here. Data fabric is an actual form of data architecture.

Rob Collie (06:33): Yes.

Justin Mannhardt (06:34): They called this thing Fabric and there's another form of data architecture called data mesh. Inside of Fabric is a feature called Domains that is a data mesh. So everybody's just confused.

Rob Collie (06:48): This is an important part of their strategy. They're like, "Look, when we launched Power BI, which turned out to be a huge success, we also completely recast the word dashboard."

Justin Mannhardt (06:58): Yes.

Rob Collie (06:58): What everyone thinks of dashboards is now reports. And then, this dashboard thing. What's that feature? That's the thing you don't use. It's the thing that everybody is looking for, but that's not the thing that you use. You use reports.

(07:10): They're just like, "You know what? In order for this to be a success, we need something similarly completely confusing. Let's do Fabric."

Justin Mannhardt (07:21): I do agree with your assessment that, and they've said as much, they're leaning so far into bringing everything into ... I'll just say the citizen developer realm and have no intentions of slowing down. I think putting everything under this one umbrella with Fabric was actually really smart, because I don't think there's another product like that.

(07:47): You could look at something like a Databricks and you've got a lot of features that are similar over there, but there's a lot of things you don't have. Or a Snowflake. The fact that they put, let's just say, data storage preparation in the same box as consumption of analytics and action on those analytics ... That was really smart and really cool. I would've never have thought about it, so I'm excited about it.

Rob Collie (08:11): End to end. We're P3. We are very excited about this.

Justin Mannhardt (08:17): Totally.

Rob Collie (08:17): We wouldn't exist without the existing wave of citizen developer tools. Power BI, without it, all of our people would still be slinging Excel.

Justin Mannhardt (08:29): Or worse.

Rob Collie (08:30): They'd be forced to use tool X, Y, Z as an export source for Excel, so they'd be using Excel and ... For Microsoft to declare that their mission is to open the floodgates to our kind of people for everything, that's just amazing.

(08:51): Thank you, Microsoft. Thank you. Thank you, thank you, thank you. It's almost as if they were like, "What can we do for those P3 folks? What can we do for their business model?"

Justin Mannhardt (08:58): I'm sure that was agenda item number one at the CAC meeting.

Rob Collie (09:02): As it should be, but we know that's not the case. They just know that this is a place that they can go, that no one can follow them. None of their competitors can follow, and it's just devastatingly effective in the real world. They really only think about enterprise. We get to slipstream on that and we have enterprise clients, but we mostly have mid-market clients and departmental clients.

(09:22): For mid-market and departmental level deployments, everything in their platform is just a steal. The price of the software almost rounds to free. It's crazy. It's like the greatest gift to the mid-market and departmental level ever. Now, they're just expanding that. Again, this is all in Microsoft's hunt to win the enterprise battle. It's so good for the smaller orgs.

Justin Mannhardt (09:49): Not only does it round close to free, but theoretically, Microsoft could have developed a product that rounds close to free, but was also very difficult for the common user to do anything with. Sure, there's still parts about Fabric that are, I'll just say, more technical or more advanced.

(10:09): But if you learn 20% about power query and data modeling, and now with OneLake, all these things. You can do so much. These mid-market companies where they just assumed, "I'm just never going to be able to do that fancy big data stuff." Well, now you totally can.

Rob Collie (10:28): In the same way that Power BI brought ... It's a very technical sounding thing. OLAP, blah, blah, blah, blah, blah. But in terms of what it actually means for you, in plain English, it's the thing you always should have had. The ability to splice an end-to-end view across all of your different systems, all of your systems that are collecting data.

(10:50): Looking at each one in isolation tells you something, but really, even the most basic metrics like profitability or margin or utilization, whatever, it's built into their definitions that you've got multiple systems powering it. And then, to be able to change up your question, change up your view, without a bunch of engineering work or spreadsheet slogging on the fly to answer a different question across that same splice dataset?

(11:21): That's kind of like table stakes and the world never had it. What did Power BI do for enterprises? It sped up their time to market on a particular solution. It did way more than that.

Justin Mannhardt (11:32): Sure.

Rob Collie (11:33): It also greatly expanded the number of projects that they could tackle at a time. And it also increased the quality of those projects, their end impact, tremendously, but they were already technically in the game.

Justin Mannhardt (11:47): They had the money, the expertise, the resources.

Rob Collie (11:51): Whereas, for the mid-market and departmental level organizations, it's the difference between zero ... Not just one. It's the difference from zero to like 100. It's just crazy.

(12:01): So I completely agree that Fabric is going to do even more for the departmental or mid-market. It's going to do even more for them than it's going to do for enterprise. And it's going to do a lot for enterprise.

Justin Mannhardt (12:14): I have fleeting memories of this when multidimensional was transitioning over to OLPA, and you would still see someone come out, "But it doesn't do this." They'd show you the XML markup of a model and, "Look at this pointing in this feature here." Eventually, that noise goes away and OLAP reigns supreme.

Rob Collie (12:33): You mean SSAS tabular? It was always OLAP, but I know what you're saying. You're saying tabular reigns supreme.

Justin Mannhardt (12:38): Yeah, and some of that is happening with Fabric now.

Rob Collie (12:40): Already?

Justin Mannhardt (12:41): Already. It's like, "But it doesn't ..." Those things are not worth getting excited about, because you know it's going to get solved. You know it's not going to be a concern in the future, because the totality of the platform altogether is so much better.

Rob Collie (12:56): And there's so many different ways in which those little objections get solved. One of them is it just takes some time. "It's on the roadmap. Everybody relax," and it gets done. Another way that it happens is that it wasn't on the roadmap, but something else that they do in the platform makes you not even need that thing anymore.

Justin Mannhardt (13:14): You don't care.

Rob Collie (13:16): You just don't care. The river doesn't even go through there anymore. You don't need a gas station for your boat there anymore. Nope. And the third way that it gets addressed is it just gets drowned by all the positives elsewhere.

Justin Mannhardt (13:32): Yes.

Rob Collie (13:32): You just don't care anymore. So many ways. And that's what happened. I remember many years ago, sitting down to a dinner with Marco and Alberto and asking them ...

Justin Mannhardt (13:43): The Italians.

Rob Collie (13:44): They didn't even let me finish the question. They knew what I was going to ask. This was in 2015, 2016. This is a long time ago. I asked them, "Hey. Right now, if you're starting a brand new project and the client has no preference for old versus new ..." They're like, "We're going to use tabular."

(14:01): They just finished the sentence, "We're going to use tabular. Without a doubt." And that was seven years ago. There's so many more improvements made. Tabular now has capabilities that no one would've even thought to ask for in multidimensional. You wouldn't have even understood the request. That's all coming.

(14:17): To get specific, the one thing I'm sort of hanging my hat on about Fabric, other than it isn't anything, it's an umbrella term ... What it really is, is OneLake. That's not the only thing, but that's the center of everything.

Justin Mannhardt (14:31): It's foundational. For sure.

Rob Collie (14:33): First of all, we should mention we did a webinar. It's already so 15 minutes ago. We did this webinar. I still think it's really good. We also have a Fabric FAQ, and the webinar is available to watch. We'll put the links to those in the show notes. Give us the quick rundown on OneLake.

Justin Mannhardt (14:52): OneLake, most simply, is the standard storage layer for everything in Fabric. So if you think about a traditional data architecture. You've got source databases. You might hear people describe something like a staging layer. You might hear something like a data warehouse, or a data lake, and these are all separate things. OneLake brings all of that together.

(15:21): Now, all of the workloads that you have for your data and analytics, whether that's reporting or data science or ad hoc queries or whatever that might be, they now all share the same foundational data and the same copy of that data. We talk a lot about plumbing here at P3. Plumbing just got a lot simpler.

(15:43): And so, now the ease at which we can collaborate, share information across our organization, benefit from the work of others ... That's so much easier in Fabric. I don't know how many times on a project I've run into something where they're like, "Rob's got some really interesting data over there." I'm like, "Yeah, but it's in Rob's database." Now, I need to hook something up to Rob's database.

Rob Collie (16:07): Rob doesn't want you hooking up to his database. Rob is like the bridge troll who says, "No. You don't get to cross this." The thing that really started to light up for me was a Power BI model today is so often the single best source of information about something important in your company.

Justin Mannhardt (16:25): Right.

Rob Collie (16:26): A lot of effort went into developing it. Not as much effort as what would've gone into the multidimensional in the past, but a lot of incremental effort over time. That's just to improve this thing immensely. It's always evolving like the Ship of Theseus. And it does. It spans across multiple systems. It encodes important business logic, so that it tells the truth, not just producing numbers.

(16:48): It tells numbers that are actually meaningful and trustworthy. Pre-Fabric, pre-OneLake, the only endpoint you could have for that investment essentially is a Power BI report. It's like the only thing. Now, it wasn't because, well, that's the only place that makes sense to have that across the company, multi-system, cross-silo, business logic, honest view ... Your AI would want that.

Justin Mannhardt (17:15): Right.

Rob Collie (17:16): What else would want that?

Justin Mannhardt (17:18): Interesting tie-in. Microsoft finally announced the release plan for Fabric. We had an opportunity to participate in the private preview, and we kind of see ... But now, it's out in the open, what's coming down the pipe. Building on this idea of OneLake and the fact that we went from a one destination to a many destination opportunity for our data.

(17:38): Some things that were announced recently is a feature that allows me to query a data model from a data engineering notebook. So I can use Python to read a data model, use measures and things like that, and the results of some other activity I'm doing, which is really cool. With that same feature, I can also ... This is a big feature ask. For a long time. I can write SQL on top of a data model, which is cool.

(18:07): They also recently announced a feature that we're excited about called Data Activator. That's now finally in public preview. And on the roadmap for this quarter, the current state of Data Activator is I could basically pick a dimension, let's say, employees ... So I've got lots of employees. Rob, Luke. I could say, "Tell me when something changes about one of them."

(18:28): But now, I can use what's called Metrics-first Triggers is coming. I can just say, "Hey, when this measure, irregardless of its dimensionality goes off ..." And now, I'm going to be able to introduce to that also sensitivity, when it goes up or down by a certain amount.

(18:44): Again, all these things that exist in the data model, I can now feed to AI applications. I can feed off to ad hoc applications. I can integrate it back to Power Apps. There's a link from OneLake over to Power Apps now. And so, it just simplifies that whole integration component.

Rob Collie (19:03): While we're on Activator, can I mention something that's actually a bit of a downer?

Justin Mannhardt (19:08): You can.

Rob Collie (19:08): We're not all positivity here. We bring a balanced view to things.

Justin Mannhardt (19:13): Balanced politics.

Rob Collie (19:15): Yes. I looked into it, and cross-dimensional intersections are not currently on their roadmap.

Justin Mannhardt (19:23): That's right.

Rob Collie (19:24): And that, ladies and gentlemen, is a miss. They don't understand it's a miss yet. It wasn't clear to them, I don't think, why it was even valuable. I actually, in these situations, where I'm looking at things that have fluctuated ... Most of the time, I'm interested in the intersection of at least two dimensions. More often than not.

Justin Mannhardt (19:43): Right.

Rob Collie (19:44): It's not like any particular customer. For example, it's customer times products. A particular customer stopped buying a particular product, or started buying a lot more of it, or whatever.

(19:57): In our advertising reporting, it almost doesn't mean anything if a particular ad starts to perform differently. It's always performing differently, but the intersection of an ad and a keyword, it's no longer performing as well on this keyword as it used to.

(20:11): That's crucial. And I have all kinds of reports that are set up like this. I use them every day. And it's like, "Nope. No activator for you." They'll figure it out eventually. This isn't one of those cases of me saying, "Hey, this thing doesn't do the thing that the old thing did." No, no. Nothing has ever done this. I'm not going backwards.

Justin Mannhardt (20:29): I have a tendency to get in a particular trap sometimes with customers. Let's say I'm doing a training or whatever it is, "Can Power BI do X, Y, and Z?" And I'll go, "No, it can't really do that yet. Here's the workaround." This feels like one of those to me.

(20:47): I was just thinking we could go through, "How would you have done that intersectional analysis before? How would you do it today?" Because I think we got a little help. And then, as soon as we finish all that exercise explaining the work around, "Here it is. Go ahead."

Rob Collie (21:02): Circling back to why OneLake is such a big deal. In the future, with Fabric, when you're building a data model, you're not building it off in this isolated corner that only Power BI can see. You're building it in a place ... Now, it is still under the hood. Very much the same VertiPaq-style compression, high-efficiency format.

Justin Mannhardt (21:22): Totally.

Rob Collie (21:23): It's just that the Power BI format for storage has become the storage format for everything. Your measures and everything, all of the investments that you've made, the relationships and everything ... Those are also available now, again, not just to this one endpoint, which was the Power BI reports, but it's available to pipeline. It's available to SQL. It's available to whatever AI you're developing.

(21:44): As you've pointed out multiple times, most of the cost of an AI project isn't the AI, it's getting the model, the data properly structured and logicized to get it going. It's just been a shame that all this effort that's been going on in getting your Power BI models in shape has been, "No. Nothing else can have it. Not anymore."

Justin Mannhardt (22:07): If you went out and grabbed 100 data scientists and you say, "What do you spend your time on?" They'll say, "I spend most of my time wrangling the data so I could apply it." Because the machine learning space has matured so much in the last several years.

(22:21): The technology is out there. If you want to do classification, regression, sentiment, whatever you're trying to do ... That's already developed. It's really about getting your data into a state where you can actually run through those things. Now people that know Power Query can participate in that process, people that build models. There's this whole universe of people that all got more valuable.

Rob Collie (22:43): That's right. That's right. Power Query just got more valuable. Even if you don't care about Power BI, which, who are you, if you don't? But if you know Power Query, now you're an end-to-end data pipeline specialist.

Justin Mannhardt (22:57): Back before I was with P3, the company I was at, we were doing a lot of work with Power Query. I can't remember which version of SSAS, but when Power Query became the standard in SSAS was around the same time they first introduced data flows to preview.

(23:13): I remember at the time just being like, "This is game over." To see, five years later, it's a standard component. It's a first class citizen of Azure Data Factory now. That's super cool.

Rob Collie (23:27): I'm just going to put you on the spot here and say, "Justin. What are we, P3, doing about Fabric, so that we're embracing it?" It's new to our team as well. Our team is the poster child example of the type of people who are expanding their capabilities. How are we looking at it here at P3?

Justin Mannhardt (23:46): The first question I posed to our team about this, it was a simple one. It was just, "How does Fabric change everything that we do?" We definitely haven't answered all of those questions, but an example of this is we have a particular style of how we like to start a project. We call it a jumpstart. How does Fabric improve that process? How does that make us better?

(24:09): There's a dimension of, "How do we become more effective for our clients by using Fabric?" And then, the other side of that is, "How do we help unblock our clients in these scenarios?" Internally, the simplest thing we've done is we just turned the dang thing on. We've got Fabric on. We've turned on all the preview features. We've got channels and teams where people are, "Hey. I'm trying this with Data Activator," and we're learning all these things in a hive mind fashion.

(24:34): People have all these lists. You mentioned the pros and the cons. We love Data Factory and Fabric, but when this is true, Synapse is still better. For us to understand those situations is really helping us understand where the technology is at and how to engage with our clients about it. Where we're spending most of our time with customers on Fabric right now is really talking about how to get ready.

(24:59): Fabric is still in a preview state. It's not GA. It's really solid overall in my assessment, but there's still bugs. There's still questions about how we should or shouldn't do certain things. We're doing some lighthouse migrations from what a client has today over to Fabric, seeing what the potential is. We're doing a lot of ... I don't want to use the term, "Migration," necessarily, but how to fast track a customer to leveraging the benefits of Fabric.

(25:27): One of the big questions that comes up in this is, "Well, do I need to move my entire estate to Fabric?" We've got customers that have made significant investments in Synapse or Databricks or Snowflake or something. Some other thing. And that's also one of the huge benefits of OneLake is its ability to shortcut to some of these other data services and just work with them.

Rob Collie (25:49): There's some passthrough pointer thing, where you don't actually have to go through the full ingestion process, lift and shift, whatever. It's like, "No, no, no. Just point to it. It'll act essentially like it's already here."

(26:02): I'm sure there's some trade-offs there, but that gives the opportunity to achieve things incrementally instead of all at once. We launched what is now currently called the Raw Data By P3 Adaptive Steering Committee on LinkedIn.

Justin Mannhardt (26:19): Working title.

Rob Collie (26:20): A working title. However, every time I change the title, the things I tell people to search for to find it on LinkedIn, they can't find it.

Justin Mannhardt (26:26): So it's going to stick. That's what you're telling me?

Rob Collie (26:28): So in the last episode, I told people to search for Raw Data By P3 Adaptive on LinkedIn. I created the group. And then, I'm like, "Raw Data By P3 Adaptive ... That's pretty wordy. I'll just take the, "By P3 Adaptive," out. And then, I realized that all the people who are searching for it just got orphaned.

(26:43): I put that back this morning. It's definitely back to a full mouthful. But anyway, we got a couple of very specific questions. We gave a little preview that we were going to be doing this. Let me make sure we get to those. So Fred Kaffenberger, he asked, "Does Fabric mean starting from scratch for companies? If you're using Azure and Synapse, or Azure Synapse, are you starting from scratch? How useful will Purview be for governance?" That's two questions.

Justin Mannhardt (27:13): Two questions. First question. Starting from scratch? No. And the reason for that is Synapse, its underlying storage layer is also Azure Data Lake Gen2, which is the foundational tech for OneLake. Again, the ability to make that accessible to Fabric is there.

(27:34): So if you're very mature in your Synapse implementation, Synapse isn't going anywhere. There's no reason to freak out or feel like you've got a bail on it. I think there's maybe opportunity to pay attention to where you would do things in Fabric going forward as opposed to re-engineering everything, so don't feel like you've got to start over at all.

(27:53): Second question was, how does Purview come into this? What is Purview? Purview is Microsoft's solution for data governance, which includes data protection, compliance, risk mitigation, data access controls. There's a lot in Purview. Fabric is going to fall into Purview's umbrella.

Rob Collie (28:15): Wait. Wait.

Justin Mannhardt (28:16): I think the most ...

Rob Collie (28:16): Fabric is going to fall under Purview's purview?

Justin Mannhardt (28:19): Yeah. We're going to strike that.

Rob Collie (28:21): No. We're going to keep that. That is a hot dad joke. That is on-brand for this podcast. Have you been listening to this show, Justin?

Justin Mannhardt (28:32): So as an example, Purview allows us as data governance professionals or data administrators to understand things like lineage very clearly. For example, a column in a Power BI dataset, being able to see, "Where did that all come from?" From its source, through all the steps in between, what was happening to it. Then, also to apply policy across our data estate.

(28:56): For example, if I have some very sensitive data in a database somewhere, if I apply a policy to that, that says, "This data cannot be exported." I'm making just a simple analogy. That policy is going to follow that data everywhere it goes. So if someone is connecting to that database with Excel or with Power BI or some other tool, it simplifies that landscape.

(29:22): Fabric is going to fall under this. The Purview integration though, for Fabric, it's a ways out on the release plan, which is disappointing, but it's like late next year. So I think it's too early to speculate exactly what we should or shouldn't be excited about in that area.

Rob Collie (29:38): Purview certainly sounded based on your description that it was very much adjacent to, or at least conceptually, to OneSecurity.

Justin Mannhardt (29:45): I would say adjacent. So OneSecurity, I'm glad you brought this up. These features are also a ways out, which is one of those, "When are we going to benefit from that?" OneSecurity, the fundamental idea is if I secure tables in OneLake, that security is going to follow that table everywhere.

(30:08): Instead of defining security in my data warehouse, and then defining that security again in my Power BI data set, I just define it once. And then, I don't have to worry about it.

Rob Collie (30:18): If you've thought about ... Abilities can be exported as just a security permission, right?

Justin Mannhardt (30:24): Mm-hmm.

Rob Collie (30:25): Now, suddenly it's like, "Well, Justin can export, but Rob can't." Now, it's OneSecurity instead of Purview. But no, no, no, no. If it's policy, that means it's a global thing for everyone. Now, it's in Purview. There's a very fine grain semantic here.

Justin Mannhardt (30:41): And Purview is a really rich product. It covers features specific for helping companies with things like insider trading.

Rob Collie (30:49): Facilitating it?

Justin Mannhardt (30:50): Yeah. Facilitating it.

Rob Collie (30:51): Facilitating insider trading?

Justin Mannhardt (30:52): Facilitating insider trading.

Rob Collie (30:53): Make it even faster. You don't want to wait around all day for your inside info. You need to trade now.

Justin Mannhardt (31:04): Purview makes it easy. What I was going to say on OneSecurity is they did announce, I believe last week, they now have both column and row-level security and Fabric SQL endpoints that will follow that.

(31:19): They have parts of OneSecurity, but not the full thing yet. So I think that's a potential thing to be aware of when we're thinking about moving to Fabric is, "Where are we really at with those aspects?"

Rob Collie (31:30): Let's get to Sue's question, Sue Bayes. She asked, "How do you think smaller businesses will benefit and what do you think the pricings will be?" We've already talked about this at the beginning. I couldn't resist. This is huge, absolutely huge for the smaller orgs. However you want to define them.

(31:46): In the same way that Power BI was, it's the difference between being in the game and not being in the game. Ironically, as a result, you're able to move faster than the enterprises can, because the enterprises still have the downside of massive org, massive team, all that stuff. Small teams are an advantage with tech like this.

(32:04): But what about the pricing? Are we still going to have essentially the pricing that is just ridiculously good? Do we know anything about the pricing at this point? Imagine a world in which Microsoft only sold Power BI via the premium model. Now, it's multiple thousands of dollars per month.

Justin Mannhardt (32:19): That's right.

Rob Collie (32:20): Just to get in the game, which still isn't that bad, depending upon how large your org is. But you could imagine a world where they switch to something like that, and that would be quote, unquote, "Bad," for the smaller orgs. It would still be amazing, but it wouldn't be rounding to free like it is at the $10 per user number.

Justin Mannhardt (32:37): So spoiler alert. We know the answer to this.

Rob Collie (32:41): Do we? I don't.

Justin Mannhardt (32:42): Allow me to be the first to tell you and enlighten you on Fabric's pricing. So Fabric is sold as a capacity model. For comparison, today, there's something called a Power BI Pro license, which retails at $10 a month per user.

Rob Collie (33:01): Yep.

Justin Mannhardt (33:01): There is not a Fabric per user model.

Rob Collie (33:05): Okay. So it is more like the premium model.

Justin Mannhardt (33:08): Now, what's different is the range of capacities is quite a bit larger than premium.

Rob Collie (33:17): Wider. Wider range.

Justin Mannhardt (33:18): Wider range.

Rob Collie (33:19): Does that mean it goes lower?

Justin Mannhardt (33:21): It goes lower.

Rob Collie (33:22): How much lower?

Justin Mannhardt (33:23): For comparison, the cheapest premium capacity on the market today is about $5,000 for Power BI. So $5,000 a month. The cheapest Fabric capacity is $260 a month. And there's another layer of this, which is there's going to be two pricing models. One's going to be pay as you go. You basically pay by the hour, so you can turn these things on and off.

Rob Collie (33:48): AKA, the surprise bill.

Justin Mannhardt (33:50): The surprise bill. Yep. "Sorry, Rob, I didn't mean to leave that big database on."

Rob Collie (33:56): I left the refrigerator running with the door open and now ...

Justin Mannhardt (34:01): So there's going to be a pay as you go model. And then, there's going to be what's called a reserved instance model. This is maybe going to be for mid-market or enterprises that just know they're going to use this. You basically pay a year or a couple of years at a time, and you get a pretty hefty discount for that.

Rob Collie (34:17): Okay.

Justin Mannhardt (34:18): One of the things we can help with too here is, "Well, what capacity do I need?" Because this is kind of Greek. It's this thing called a capacity unit. It's some arithmetic around CPU and RAM, but it's like, "What does it really mean? What do I really need?"

Rob Collie (34:33): If your capacity fluctuates a lot like month to month, does that mean we're using the flux capacitor? I know. I know. This is ...

Justin Mannhardt (34:46): That's a much better name for Autoscale.

Rob Collie (34:48): Auto Flux Capacitor. We should just call it that. All right. We're up against our hard stop here. This is good. It allows us to produce an occasional more snackable episode, but to make sure to call this out ... We really, really, really, Fred and Sue, really appreciate your questions.

(35:05): And if you're listening to this and you would also like to have your questions answered, please, please, please look us up on the LinkedIn group. You can search for Raw Data By P3 Adaptive. Join us up. We'll chat. It's been a lot of fun, even just in the early going, just the little bit of interaction we've been doing in there. I like it. It's a conversation.

Speaker 3 (35:19): Thanks for listening to the Raw Data By P3 Adaptive podcast. Let the experts at P3 Adaptive help your business. Just go to p3adaptive.com. Have a data day.