Latest / Microsoft Cloud IT Pro Podcast / Episode 336 – S Tier Microsoft Build 2023 Announcements
Transcript
- 0:03Welcome to episode 336 of the Microsoft Cloud IT Pro podcast
- 0:08recorded live on May 31st, 2023.
- 0:11This is a show about Microsoft 365 and Azure from the perspective of it pros and
- 0:16end users where we discuss the topic or recent news and how it relates to you.
- 0:21Microsoft build has come and gone and we have some updates to discuss this week
- 0:25regarding announcements made at the conference. As can be expected,
- 0:28there was a bunch of AI related news around co-pilot chat G P T and OpenAI
- 0:33and then finally a new SaaS-based data product, Microsoft Fabric.
- 0:40Are you gonna catch me up because I was on vacation last week and you would be
- 0:45proud of me Scott.
- 0:46I turned on my computer I think once the entire week cuz
- 0:51I actually had to do some podcast stuff.
- 0:54I have tried to catch up on blog posts and news and all of that,
- 0:57but I have not had the time I wanted to to dive in to build stuff and I saw
- 1:02you put stuff and I've flagged some articles I saw but again,
- 1:07I have not gone into depth on anything in particular.
- 1:11So what shiny build stuff should we talk about? So let.
- 1:15Me catch you up on some Microsoft build stuff.
- 1:18So Microsoft Build 2023 happened last week, two days.
- 1:23You can absolutely go out and watch the on-demand sessions at this point.
- 1:27It's all out there ready to go. All the blog posts and things,
- 1:31we'll have links in the show notes and all that good stuff.
- 1:35So I highly recommend everyone go watch the keynotes
- 1:40even though keynotes are largely useless.
- 1:43I think it's very interesting to just see the kind of tone and approach
- 1:48around things like ai. I was,
- 1:51I was kind of joking around as the keynote was going on.
- 1:53Like my big takeaway from the keynote was that you're going to be able to build
- 1:58co-pilots for your co-pilots to pilot your co-pilots,
- 2:02to get you another co-pilot to do something for you. And,
- 2:06and I was very confused by this because I had been introduced and
- 2:11I think most of us has been introduced to co-pilots as marketing terms
- 2:16and not as a,
- 2:18a feature like it's a product versus a feature like GitHub co-pilot
- 2:22is, you know, GitHub's ability to do AI generated,
- 2:27ML generated code for you M 365 co-pilot like copilot for
- 2:32Word is the same thing,
- 2:33it's a large language model that can respond to prompts in Microsoft Word
- 2:38and Microsoft kind of came out and they said sure we have copilot
- 2:43and it's a marketing term so it's,
- 2:46it's a product and a feature and now we're gonna let you go build your own
- 2:51co-pilots as well. So not only are there the existing co-pilots,
- 2:56but theoretically you could build a co-pilot for another co-pilot.
- 3:01Like say you wanted to, I don't know,
- 3:04you wanted to build a co-pilot for Word that interacted with Microsoft's
- 3:09co-pilot. Like that would theoretically be possible.
- 3:11Like you could do some own pro,
- 3:12some of your own prompt engineering and prompt hacking and inject on
- 3:17top of their L L M with your L L M and and potential
- 3:22responses or context of your business and and things like that on top of it.
- 3:26So that was kind of interesting to go through and see.
- 3:29I also recommend that everybody watch the keynote that happened right after
- 3:33that, which I believe was with Kevin Scott,
- 3:36he's the recently crowned VP of AI or something
- 3:41like that at Microsoft. I forget his ex exact title,
- 3:44but he kind of goes into more of that ecosystem and everything that's going on.
- 3:49So a couple things, uh, co-pilots for everybody.
- 3:51Like absolutely you can go out and build your own co-pilots.
- 3:54There's a service inside of Azure called Azure AI
- 3:59Studio where you can go and train these models and design your
- 4:04co-pilots and release them out there.
- 4:06Some of the demoware I saw was kind of,
- 4:09it was a kin to the old like chatbot in five minute kind of demos that you
- 4:14used to be able to do.
- 4:15I think it's a little richer than that just based on kind of the language models
- 4:19and and what they have access to today and just kind of how far prompt
- 4:23hacking has has come along or I guess prompt engineering to get things to where
- 4:27they need to be.
- 4:28The other interesting thing that was announced was
- 4:33that plugins are coming to chat G P T integration in
- 4:38Bing. So today bing kind of runs on the,
- 4:42I think the G P T three five turbo model and it is coming over
- 4:47to G P T four and when it does that,
- 4:52it will have access to the same exact plugins that you have access to if
- 4:57you are a premium subscriber open AI today.
- 5:01So if you're already paying open AI,
- 5:02the monthly fee for access to the paid version of that product,
- 5:06you can already flip over to the G P T four model and with
- 5:11G P T four you get plugins.
- 5:13So plugins will let you do things like have more context around
- 5:18kind of these outside data ecosystems.
- 5:21So some of the examples that they brought up cuz it's, it's bing,
- 5:24it's kind of consumer grade and and consumer wear. Yep.
- 5:26Where things like open table Zillow,
- 5:29you know like hey I want you to go out and build me a table of
- 5:34houses that are within five mile radius of this address that meet these
- 5:39criteria and put them in a table and rank them by price and number of
- 5:44bedrooms, number of bathrooms, things like that.
- 5:47So plugins are coming and and rapidly coming and I'm kind of looking forward
- 5:52to those. I think they'll be nifty especially once you know,
- 5:56some more stuff comes out there like I would love to see a plugin for something
- 6:00like Reddit like that's built around the Reddit API and not just bing
- 6:04scraping Reddit and it has context of things like upvotes and kind
- 6:09of hot articles and and things like that.
- 6:11I think it's gonna make data mining with some of the LLMs just a
- 6:16lot smoother at the end of the day.
- 6:18How do these plug-ins work then? Where do they live? Is this again,
- 6:22I haven't read up on this at all.
- 6:25Is this plug-ins to connect to Bing two services or plug-ins that you can put
- 6:29like in Bing for instance you said build me a table of houses
- 6:34on Zillow. Does that return,
- 6:37is that a plugin for Bing to go out to Zillow and then it just returns on like
- 6:42the Bing chat UI or how,
- 6:45I guess how do those plugins work and where do they sit? Does that make sense?
- 6:50Yes, so the plugins, my understanding of them,
- 6:54and I haven't written one of these so don't like hold my feet to the fire.
- 6:57My understanding is plugins are executed within
- 7:03the context of chat G P T but then they basically have
- 7:07connectors to go out and be able to interact with those other
- 7:12services.
- 7:13So you can almost think about it as like a plugin is exposing another API
- 7:18to chat G P T so that it can go out and interact with that data and then that
- 7:23data's returned in a format and in a manner that is all up friendly
- 7:28to chat G P T and and being able to have everything just kind of come back
- 7:33and be able to be massaged into the right format. So you're likely,
- 7:37it's not like,
- 7:38I don't think it's like Bing has gone out and indexed all of Zillow,
- 7:43it's more like it's calling Zillow with a pre-engineered prompt and then
- 7:48Zillow has an a p I endpoint that's able to respond to that prompt based on
- 7:53either a data set that they've built out or maybe they even call into another L
- 7:58L M on their side to get some data back from it and then they return all that to
- 8:02the client munt together and it's all kind of ready to go.
- 8:05Okay. I think that's making sense and I'm still trying, it's.
- 8:09Basically like APIs, like it is just API interaction.
- 8:14I think the interesting thing is it's almost like APIs talking to APIs just
- 8:18because of the human language nature of L L M prompts and kind of how,
- 8:23how they're constructed, right? It's it's,
- 8:25it's not like you are writing a database statement and you're saying you know,
- 8:30set options strict on,
- 8:32you're going in just human language and you're saying hey large language model,
- 8:37I need you to imagine that you are a CEO
- 8:42at blah blah blah blah company and you always act in this manner and you're
- 8:46gonna be kind and you're gonna respond in this way and you're gonna be held to
- 8:50this set of standards in your response. You know,
- 8:52where I think we're used to more like just arbitrary like binary on off kinds of
- 8:57things as we're, you know,
- 8:58maybe scripting or or interacting with programs today or or writing our own.
- 9:03So it's a little more free flowing than that. So I, I don't know,
- 9:07we'll see what comes out of it. Like.
- 9:08Developers are writing these that and essentially submitting these two yes G P
- 9:13T or Bing and saying hey here's my plugin.
- 9:16I want to bring in this additional data or add this additional functionality
- 9:21to G P T. So is this adding it to chat G P T?
- 9:25Is this adding it to open ai? Is it adding it like this is also,
- 9:28I was starting to look through this,
- 9:30this is getting all kind of confusing because everybody kind of uses chat G P T
- 9:34as a default but it looks like open AI is now using Bing as the
- 9:39search experience for chat G P T but I don't think Bing Chat is
- 9:43necessarily chat G P T and trying to keep all the terminology straight in my
- 9:47head too. So.
- 9:48Open AI is the company, they're the big broad company. They build a bunch of.
- 9:53Models and the learning models and all of that. Yep, yep.
- 9:56They build a bunch of products.
- 9:58One of those products is called Chat G P T and then within
- 10:03chat G P T you have various flavors of language
- 10:08models that are available to you basically like what's the switch?
- 10:12I wanna flip and chat G P T for the types of responses that I wanna give back.
- 10:16Those are all the G P T models you hear about.
- 10:19So you hear about like G P T three five, G P T three five Turbo,
- 10:23G P T four, those are kind of the features.
- 10:26Right? Built by open ai right?
- 10:28And then chat G P T is sitting on top of those different flavors of G P T
- 10:323 3 54 chat.
- 10:34G P T is just branding like chat G P T is a website.
- 10:37It's the thing you go to to interact with a set of features that you turn on or
- 10:41off and the models are the features that you go back and forth and you turn on
- 10:46and off that way.
- 10:47So you could go in like if you're a paid subscriber to chat G P T today,
- 10:51you can go into chat G P T and you can say okay I wanna work with the
- 10:56G P T three five Turbo model and the G P T three five turbo model has
- 11:01a certain set of constructs.
- 11:03It was built on top of one Lang large language model.
- 11:06Or you can say I wanna do G P T four And probably the biggest difference between
- 11:10like some of the older models and G P T four is that G P
- 11:15T four is the one that has the plugins and it can do like real live internet
- 11:20searches.
- 11:21So remember like when you chat G P T initially came out and they said okay we
- 11:24have this model right that you can interact with and don't worry about it.
- 11:27One of the big constraints of it was that model was only trained on a data set
- 11:32that was current up to late 2021. Yeah.
- 11:36I remember there was a date there, I couldn't remember if it was 2020 and 2021,
- 11:39but I do remember that where you'd go ask it stuff and it'd be like my data's
- 11:43limited to data up to this point in time because it,
- 11:47it couldn't go grab recent news.
- 11:49Recent articles didn't really touch the internet. Correct.
- 11:53So G P T four opens that up,
- 11:56lets it go and do more realtime contextual searches and
- 12:01the Bing part of it is Bing is using ,
- 12:05the G P T four model or transitioning to it cuz they will be as they bring like
- 12:10plugin integrations and things like that. Yep.
- 12:13So Bing is using a large language model in the background and then the
- 12:18background of GP chat G P T, they're also using Bing as their search engine.
- 12:22So what they've said is like G P T four when you're using that model then it
- 12:26needs to reach out to the internet and it needs to do a search in real time.
- 12:29It's gonna use Bing as the search engine of choice to do that.
- 12:31Got it.
- 12:32So that's how they're kind of connecting G P T four and to be able to search
- 12:36some of the internet as leveraging Bing.
- 12:38But then Bing is using G P T four in the background as well.
- 12:43It's a.
- 12:43Little weird. So Bing would be the backend search engine for chat G P T. Yep.
- 12:48And chat G P T would be the backend large language model for Bing. Okay.
- 12:53You could almost like draw like a weird circle and put some double arrows on it
- 12:57and it, it would make for you know, a pretty funky physio diagram. That's.
- 13:00What I was trying to kind of wrap my head around as we were talking through this
- 13:03is how all of those were tied together but what you just explained makes sense
- 13:08and then they're using all these same standards then for plug-ins that if you
- 13:12write a plug-in for open AI because of the weird circle model,
- 13:15all those plug-ins will work in chat G P T and they'll work in Bing and
- 13:20they'll work in co-pilot and all of that because it's all just one big happy
- 13:24circular family. Correct.
- 13:25Like theoretically that's, yeah that's, that's the way it all goes.
- 13:29So we'll see if it manifests that way but .
- 13:32And then you can go build your own co-pilot on top of all that using those
- 13:36add-ons and yeah.
- 13:37You can.
- 13:37So you can go out into Azure AI studio today and you can do that another place
- 13:42that co-pilot is coming in the Microsoft ecosystem that I
- 13:47thought was kind of interesting as I spend more and more time there is Power bi.
- 13:52So there's gonna be a Power BI desktop developer mode
- 13:56that's going to enable a workflow for interacting with,
- 14:01well I guess a couple things.
- 14:02So you're gonna get co-pilot integration on Power BI and then you're also going
- 14:07to get integration and some other things coming to Power BI that I thought were
- 14:11kind of exciting. Like I, I would totally use that all the time.
- 14:14I'm tired of sending the same P B I X file to 30 people and then wondering where
- 14:18they've uploaded it to.
- 14:22Do you feel overwhelmed by trying to manage your Office 365 environment?
- 14:26Are you facing unexpected issues that disrupt your company's productivity?
- 14:30Intelligent is here to help much like you take your car to the mechanic that has
- 14:34specialized knowledge on how to best keep your car running Intelligent helps you
- 14:38with your Microsoft Cloud environment because that's their expertise.
- 14:42Intelligent keeps up with the latest updates on the Microsoft Cloud to help keep
- 14:46your business running smoothly and ahead of the curve.
- 14:48Whether you are a small organization with just just a few users up to an
- 14:52organization of several thousand employees they want to partner with you to
- 14:56implement and administer your Microsoft Cloud technology,
- 15:00visit them at intelligent.com/podcast.
- 15:04That's I NT E L L I G I N
- 15:08k.com/podcast For more information or to schedule a 30
- 15:13minute call to get started with them today.
- 15:16Remember Intelligent focuses on the Microsoft cloud so you can focus on your
- 15:20business. Oh co-pilot everywhere,
- 15:25you know where else co-pilot's coming everywhere to Edge .
- 15:29Did.
- 15:29You see co-pilot as coming to Edge too? Yes. So this one could be interesting.
- 15:33This one of some of them intrigues me a little bit where it essentially uses the
- 15:38website as context. So some of these news articles or if you pull up a website,
- 15:43this one would be interesting to play with co-pilot to get like a summary of a
- 15:47webpage or to find out more about the webpage.
- 15:51This one is one of the co-pilots that intrigues me more.
- 15:54I haven't played with like Power Apps or Power Automate or some of those
- 15:57co-pilots.
- 15:58I'm really looking forward to getting my hands on all the co-pilots and seeing
- 16:01which ones I actually like and I find useful and which ones are just kind of eh.
- 16:05So for Power Automate I haven't used it much but I'm not
- 16:11a large power automate person.
- 16:13Like I'm not using it in my daily life but I do use Power
- 16:18BI almost daily and okay, I've got Power BI desktop open all the time.
- 16:22So there's a quick little video preview.
- 16:24It's actually on the news site from like Microsoft Global Comms.
- 16:28I'll put a link in the show notes to that which is copilot in Power BI
- 16:34and you can kind of see how it works.
- 16:36So one of the interesting things to me about copilot and Power BI is you know
- 16:40how Power BI has kind of the q and a widgets and things available
- 16:45today? Yeah. Where you can go in and it understands your data set and so,
- 16:48you know, show me sales by quarter for the past two years ranked by
- 16:54industry of company, blah blah blah or industry of customer. Uh,
- 16:57and and it can do all that.
- 16:59So that already exists today but all that's doing is building out like singular
- 17:04visualizations for you.
- 17:06The co-pilot in Power BI lets you go in and it says hey here's a
- 17:11data set, like here's a database and here's some tables in it.
- 17:15Like maybe go figure that out for me and then tell me what I can do with it.
- 17:19Oh and by the way go ahead and actually build out the P B I.
- 17:23For me it's kind of like I,
- 17:25I get it's all demoware and when it works it works but it's super
- 17:30cool just to kind of like see it just, you know,
- 17:33get released and go against things like that. Cause if you think about it,
- 17:36like lots of companies are the same at the end of the day.
- 17:40Like if you're a publicly traded company,
- 17:43like the way you're keeping the books is gonna be very similar to all other
- 17:47publicly traded companies usually in your country cuz that's some type of
- 17:50regulated industry.
- 17:51So there's gonna be certain requirements for you for reporting your data,
- 17:54for storing your data.
- 17:55Like it's always gonna have to come out in a common format anyway.
- 17:58So it's one of those things that you can absolutely train a model on and then
- 18:02have that model come back and kind of immediately have context to your
- 18:06data but it already knows things like common schema, common terminology, all,
- 18:11all that kind of stuff.
- 18:13Interesting. This is, I'm Scott,
- 18:15I'm committing to not adding everything to my list.
- 18:17This one I don't use Power bi enough.
- 18:19This is not gonna get added to my list but it would be interesting and maybe if
- 18:24I started pulling more data into it, yeah, I'll have to play with it.
- 18:27Maybe this is a maybe how about it? Maybe on my list, maybe not.
- 18:31Uh, maybe on your list. Okay.
- 18:33So maybe one last one for today before we run out of time,
- 18:36cuz I think we both have meetings coming up.
- 18:39And this was in the same article with introducing co-pilot and Power BI
- 18:44was introducing Microsoft Fabric.
- 18:48I have seen a boatload of headlines about Microsoft Fabric.
- 18:53I am not a data guy and this looks
- 18:57like it's tied in a lot more to Synapse data,
- 19:01factory data warehousing data science, that type of stuff from the headlines.
- 19:05And what I've gathered is that it's like this cross
- 19:10cloud data ingestion thing where I can like go grab all my
- 19:15data from all the things everywhere and pull it into Microsoft Fabric and
- 19:19integrate it into some of these data analytics tools into Synapse Data
- 19:24Factory. Is that right? Or like what,
- 19:27what is fabric at a high level?
- 19:29Because I also don't know that we have time to dive into it in detail.
- 19:32I don't know , I don't know enough about Fabric. All.
- 19:35Right. Fabric. Well first of all it's a horribly named product.
- 19:39I think some marketers fell asleep at the Wheel.
- 19:42again.
- 19:43When they came up with the name for it. Y yeah.
- 19:45Cause there there are already things called fabric in the Azure ecosystem
- 19:51and and in the broader Microsoft ecosystem as well, there's.
- 19:54Lots of fabric. Like my first thought was App Fabric.
- 19:57I can't remember what App Fabric was from. Was that a SharePoint thing?
- 20:00Well it was Azure. There's App fabrics. Service fabrics, yeah, yeah,
- 20:04there there's fabrics for service fabrics.
- 20:06All the things.
- 20:07So Microsoft Fabric is a
- 20:12SAS service.
- 20:13So it's kind of interesting in that it is something
- 20:18that wraps a bunch of Azure services while not
- 20:23necessarily being Azure itself.
- 20:28What I mean by that is if you think about some of these things, like we were,
- 20:32we were talking about Power BI like a couple minutes ago.
- 20:34Power BI is a SaaS service. It's licensed by the user.
- 20:39Like you don't go out and buy a Power BI tenant and then
- 20:44buy a Power BI database and then kind of put all these pieces together yourself.
- 20:48You say like, oh no,
- 20:49I'm just gonna buy a per user license to this thing called Power bi and it's
- 20:53gonna let me do visualizations based on a bunch of data sources that exist out
- 20:57there. So it's,
- 20:58it's a little bit of a weird model for me to wrap my head around,
- 21:00I guess is just like data as a service that's not
- 21:05necessarily Azure focused.
- 21:08It's like it's Microsoft focused. So it's not even M 365,
- 21:12like it's a whole new weird thing. But at a high level it's,
- 21:16it's a SaaS offering that brings together a bunch of building blocks for
- 21:21you inside of a single service offering.
- 21:25So while you have all these component pieces and parts like Power BI and
- 21:29Synapse and Data Factory and Azure Data Explorer and all these things
- 21:34that sit out there, you know, they are disjointed.
- 21:37So you could go down the path of picking up a Power BI license for your users
- 21:40and then saying, okay,
- 21:41I'm gonna use my existing Azure data explorer thing over here.
- 21:44But then what do you do when you want to go grab data from like a raw data set
- 21:49from a data lake and transform it with Spark? Okay,
- 21:51well now I gotta go create a,
- 21:53a storage account and activate hierarchical namespace on it and then,
- 21:57oh by the way, I might need like a, a synapse spark pool to spin that up and,
- 22:02and you just end up with all these pieces that you need to manage and it might
- 22:06be a little bit easier to have it in a managed offering,
- 22:08which is kind of what Fabric does.
- 22:10So it brings together Data factory synapse, data engineering,
- 22:14synapse data warehouse, synapse data science, synapse real-time analytics,
- 22:18Azure Data Explorer, power bi,
- 22:22and it brings together cloud storage as well
- 22:27and even multi-cloud storage into a single
- 22:32service offering. So now rather than saying, Hey user,
- 22:36go to Power BI and then go talk to the developer over here to get a connection
- 22:40to that data lake and then go talk to the developer over there to get a
- 22:43connection to their, their data set. And then, oh by the way,
- 22:46go talk to this other developer in this other team who's been doing this Delta
- 22:51Lake project and blah blah blah. You can just have it all sit in one place,
- 22:56which is your one lake, which is kind of like your foundational data set.
- 23:00And then all these services automatically have connectivity and they know how
- 23:05to talk to that one lake and consume data out of it.
- 23:10So it's a little weird. Like you get all these Azure services,
- 23:13you get some M 365 or like data SaaS services like
- 23:18Power bi, then you can also pull in data from outside sources.
- 23:23Like we'll see how it goes. I, I would recommend there's a,
- 23:27in the getting started docs,
- 23:29there's a fabric terminology bit of documentation and it
- 23:34takes you through like general terms of the service. Like hey,
- 23:37what's a workspace? You know, how do they measure capacity? Things like that.
- 23:42But then it breaks down into the individual pieces and components like inside
- 23:46Data Factory inside of Microsoft Fabric.
- 23:50What do you get inside of uh,
- 23:52synapse data science inside of when it's wrapped in Microsoft Fabric,
- 23:56what do you get?
- 23:57So I can see it like making things a lot easier for people. You know,
- 24:02if you think about like, you know,
- 24:03just that headache of like getting access to a data set and getting like my VM
- 24:08and my managed identity to be able to talk to this storage account or this thing
- 24:12over here, this database over here. Like that's as painful, right?
- 24:15Like why not just have it all managed by the service and wrapped around a common
- 24:19data layer? Like that makes sense.
- 24:21So I think that's kind of like how the service manifests and comes out.
- 24:26I haven't seen,
- 24:26and I I didn't do too deep a dive on it about things like pricing
- 24:32and what that's gonna look like, like impacts to existing users.
- 24:38It's kind of like if you're a Power BI , uh, the,
- 24:40the way it is in the docs, if you go to the docs today and you say, Hey,
- 24:43I wanna start a fabric preview trial,
- 24:45it immediately opens up with what if you're an existing Power BI user.
- 24:50And I think that kind of gives you a big hint as the target audience and,
- 24:54and where they're going with it. Yeah.
- 24:56They do have,
- 24:57like I was looking through some of the documentation where they have by fabric
- 25:02now or by Pro and I don't see though, to your point,
- 25:06I don't see purchase fabric capacity. There's some pricing but man,
- 25:11some of the naming in here too, Scott,
- 25:13like I'm gonna go back to you with marketing.
- 25:15I was flipping through this article and now you have
- 25:20Data Lake but you also have one Lake data and then
- 25:25you have uh,
- 25:26there was something else in there that I saw that was named really similar to
- 25:29that.
- 25:30I think that fabric terminology site is gonna be a good site to keep handy as
- 25:35you go sort through what all of this stuff is.
- 25:38I think the big thing to keep in mind today is, like I said,
- 25:40like it is a wrapped SaaS offering.
- 25:42So you don't go and buy Data Factory by itself anymore.
- 25:46Like you or you still can like I'm, I'm not saying you can't do that,
- 25:50like absolutely like the component pieces and parts are there,
- 25:53but if you walk into it from a service offering side of things, it's just,
- 25:57I'm gonna go buy Fabric And Fabric has a data factory component within it. Yeah.
- 26:02Well and it looks like because of that, as you need,
- 26:05depending on the size of your data,
- 26:06it looks like there is different capacities and different levels and there's,
- 26:10this isn't just gonna be go buy fabric,
- 26:14but you're gonna have to buy different levels or different capacity
- 26:19of fabric based on the amount of data, how much is gonna be in there,
- 26:24et cetera, et cetera.
- 26:25Capacity is an interesting one. So it's a SaaS product,
- 26:28so it it's a multi-tenant kind of product. So you,
- 26:31you're basically buying a fabric tenant is is really what you're coming into
- 26:36when you go ahead and purchase that.
- 26:38So each fabric tenant has
- 26:43capacity that's measured in things like compute, like hey,
- 26:47I'm spinning up synapse in the background.
- 26:49What are the different SKUs that are available to me?
- 26:52How many cores is that gonna come with?
- 26:55So immediately like if you start going down the fabric licensing hole and you
- 26:58get into like capacity and SKUs,
- 27:00it opens up with compute where you've gotta figure out like what's the skew and
- 27:04fabric to the equivalent skew and Power BI and what does that translate
- 27:09into V cores and how much you use
- 27:14along the way Gets a little bit weird on that side.
- 27:17The whole OneLink concept is interesting as well because it's
- 27:22not a data lake that you manage. So I guess coming from Azure storage,
- 27:27you know my take on, it's like just bring your data lake, right?
- 27:29Like you've already got it out there, it's just a, a gen two storage account,
- 27:34GV two with hierarchical namespace enabled, blah blah blah. Yep.
- 27:36But now with one lake, because it's a multi-tenant model,
- 27:41they've effectively like provisioned some storage on the background for you.
- 27:45So I think that's gonna be interesting to see like how they just price capacity
- 27:50and and kind of figure all that out.
- 27:52And then they built connectors from things like one lake over to all
- 27:57the other clouds as well to to Microsoft's cloud and then over to S3 and
- 28:02GCP as well. So if you have a one lake,
- 28:05there's a connector in there where you can go and say like,
- 28:08Hey I have an S3 bucket that sits over here that maybe has a data set that you
- 28:13want in it and you can actually connect your one lake over to that S3 dataset
- 28:17and then the proxy layer of one lake just makes it super easy for Power BI to
- 28:22talk to it. Like Power BI doesn't need to know that it's talking to s3,
- 28:25it just needs to know that it's talking to one lake and then you're done.
- 28:28Interesting. Lots of stuff to go figure out.
- 28:32This is not what on my list either.
- 28:34This is not a high priority for me given what I do,
- 28:36but with all the news I've seen around it about it,
- 28:40I figured I should go dig into it a little bit.
- 28:42I think you're gonna see it more than you think in your world
- 28:47because your customers are Microsoft 365 SaaS customers who
- 28:52use things like Power bi. So if Power BI is an entrance point today,
- 28:58like looking at the way that it's positioned,
- 29:01it was announced and the way the documentation's oriented and things like that,
- 29:05it's kind of like Microsoft is positioning it as every Power BI customer
- 29:10should just become a fabric customer and then your life will be a lot easier,
- 29:14right?
- 29:14Because you get all the compute and it's managed and identities are managed and
- 29:18you don't need to worry about security and things like that.
- 29:20So if you in your ecosystem have customers who are doing Power bi,
- 29:24it might be something you wanna spend some more time with cuz I bet that's where
- 29:28like a bunch of the bleed outcomes from is existing Power BI customers going
- 29:32like, huh, should I just be doing Power BI anymore or should I be doing uh,
- 29:36fabric?
- 29:37I feel like it'll depend a lot on the pricing just looking through,
- 29:40I posted a link to it in Discord too,
- 29:42looking through the whole Microsoft Fabric licensing.
- 29:46It feels way more confusing than Power BI . But I guess to your point,
- 29:50if you're starting to bundle in some of those locations where your
- 29:55data might be housed,
- 29:56maybe it's that the bundle does end up being cheaper,
- 30:00the licensing does not look like it's gonna be straightforward based on that
- 30:04fabric licenses. So I'll be curious and I'm gonna keep an eye on it.
- 30:08It's gonna take time to get there because Fabric is kind of,
- 30:13cuz you're buying Rebundled capacity.
- 30:14So now you're back to like this weird model of saying like, oh, okay,
- 30:19like what's the skew I get and what's the capacity that comes with it
- 30:25along the way,
- 30:26which can get confusing pretty quickly.
- 30:30Well, with that, I should probably go jump to my next meeting. I can let you go.
- 30:34We've been chatting for a while now. All right,
- 30:37we can continue on Friday cuz our recording schedule is a bit
- 30:42on with some vacations coming up,
- 30:44but we'll talk to you again in like two days for more topics. All.
- 30:47Right, sounds good. Thanks Ben. Thanks.
- 30:48A lot Scott. We'll talk to you later. If you enjoyed the podcast,
- 30:54go leave us a five star rating in iTunes.
- 30:56It helps to get the word out so more IT pros can learn about Office 365 and
- 31:01Azure.
- 31:02If you have any questions you want us to address on the show or feedback about
- 31:06the show, feel free to reach out via our website, Twitter, or Facebook.
- 31:10Thanks again for listening and have a great day.