Latest / Microsoft Cloud IT Pro Podcast / Episode 351 – Rock me like a hurricane… or Python in Excel
Transcript
- 0:03Welcome to episode 351 of the Microsoft Cloud IT Pro
- 0:08podcast recorded live on August 30th, 2023.
- 0:12This is a show about Microsoft 365 and Azure from the perspective of it pros and
- 0:16end users where we discuss the topic where recent news and how it relates to
- 0:20you. It's hurricane time or Python time,
- 0:23or maybe the hurricane is bringing pythons. This week,
- 0:27Scott and Ben catch up on some of the latest news around power automate
- 0:32Microsoft Enterra and Microsoft Excel while they look out their windows.
- 0:36Watching Hurricane Adalia whipping the tree branches around.
- 0:41Was a dark and stormy morning. Yeah, we're back to hurricane season. Yeah.
- 0:46Which means we get to sit here and record and watch the trees gently swaying in
- 0:50the breeze.
- 0:51Yeah. I keep turning around like this. It's 'cause it's gusting. We're like,
- 0:55whoosh.
- 0:56You're looking whoosh. See, I have mine right out here.
- 0:58Coming through the backyard, so we'll see how it goes.
- 1:01I don't have to completely look that way.
- 1:03I just look out my window here right now. It's really still. But yes,
- 1:07I'm the same way watching the gusts.
- 1:09Every once in a while the trees start madly dancing in the wind. Oh,
- 1:12look at you.
- 1:13You, you've been playing with chat G b T and poetry haven't you?
- 1:16Yeah, no, I should . I should have Ja chat.
- 1:20G p t write a poetry script for the podcast. Yeah, I dunno.
- 1:25We have not, I don't know about you. We haven't gotten much rain.
- 1:27It's supposed to be windy later today as it gets a little closer to us,
- 1:32we're supposed to get 60 mile an hour winds, but really no rain,
- 1:36which I'm disappointed in because I'm cheap and I don't wanna pay for water to
- 1:41fill up the pool and we're not getting enough rain to fill the pool back up.
- 1:46Yeah, well.
- 1:47It's a hard life, but.
- 1:49Yeah. Yeah. We'll see how this one's going.
- 1:51Folks on the Gulf Coast are getting it like they're, we're we're into,
- 1:54we're into landfall time right now,
- 1:56so I I was watching the news before we hopped on,
- 1:59but once again Jacksonville gets cleared out so.
- 2:03You know who didn't though? This was interesting.
- 2:05I was talking to somebody and they were like,
- 2:07I don't know if this has ever happened before, that Valdosta,
- 2:10Georgia has a hurricane warning because of where it was coming
- 2:15in through the armpit of Florida.
- 2:17There there's like very little land between the ocean and Georgia.
- 2:22So it was like just blowing right through Florida and it was still gonna be a
- 2:26hurricane when it got up into Georgia. Yeah.
- 2:28Yeah. It, it, it happens.
- 2:30It was something climate change is real Y'all like maybe when your insurance
- 2:34company has started upping the prices,
- 2:36we should have all gotten the gotten the idea. We're.
- 2:39Just dropping you because they didn't want to insure your zip code anymore.
- 2:42There.
- 2:43There's that too.
- 2:44Not that ever happened to me and I'm bitter about it. Yeah, I need coffee.
- 2:48It was a long ways away. .
- 2:51Okay then. All right, you got coffee?
- 2:53I've got coffee.
- 2:54Let's go ahead. You've.
- 2:55Got news.
- 2:56Get into it. Where do you wanna start today?
- 2:58Where do you wanna I I don't have my articles up. Just a minute. Browser.
- 3:01I was getting my coffee. Where should we start today?
- 3:03I can guide you in or you go ahead and pick one. No, you're all set.
- 3:06It's your show. Pick one.
- 3:07Oh, let's pick one.
- 3:09Let's pick the integration of power automate telemetry data with
- 3:14Azure application insights. You.
- 3:16Went the other way. I didn't think you were gonna go that.
- 3:18Way. You didn't think I was gonna go this way?
- 3:20This one's your exciting one, huh?
- 3:22I don't know if this was exciting. This one's a little older. How's that?
- 3:25This one is like a week old at this point in time. Yeah.
- 3:27Sorry I messed y'all up. I'm jumping all over the place today.
- 3:31I don't think there's not a ton with this one.
- 3:33This one an announcement back last Thursday on
- 3:37August 24th that you can now go
- 3:42into your power automate and go set
- 3:47up certain data to come out of power automate into
- 3:51application insights.
- 3:52So this can some things like cloud or cloud flow runs and power
- 3:58automate.
- 3:58I still think they need to call 'em automation since they name changed the name
- 4:01of it.
- 4:01Because anyways triggers action level data from an
- 4:06environment to app Insights and it's really just like a click,
- 4:10click Next you go through export to application insights,
- 4:14select what you wanna export and then power automate.
- 4:16And it's really those three categories and it
- 4:21just shoves it all into insights for you into log analytics so you can go look
- 4:26through it. You can also take multiple environments into a single instance.
- 4:30So I know I have some companies that have multiple instances.
- 4:33It'll be interesting to see how much data eventually gets added.
- 4:38Like one of my peeves with Power Automate is
- 4:43I will get,
- 4:45I've built some very complex flows for certain customers and when something goes
- 4:50wrong or something errors out, it says failed in the run history.
- 4:54But if you have these flows that are running like a hundred, 200,
- 4:58300 times a day and they only load 10 instances or
- 5:0220, I can't remember how many it is,
- 5:04but it's like the last 20 runs in the UI and then you have to scroll and it's
- 5:08like the whole continuous scroll.
- 5:09Then you wait for the next 20 to load and then you scroll and wait for the next
- 5:1220 to load. It's not very easy to dive through.
- 5:18Power automate runs to view things like which actions are flowing
- 5:23and when they flow and try to dive through the
- 5:28details of what actually happened or even going and looking at a historical one
- 5:32that may be completed but didn't do what it was supposed to because that happens
- 5:37too where it doesn't actually air out but it doesn't work correctly.
- 5:41Having some additional data to be able to pull out of
- 5:46these power automate flows and really analyze the data and analyze these
- 5:50different actions and what happened with them would be very helpful.
- 5:54I don't know that this is there yet,
- 5:58but hopefully this is a step in ingesting those logs into someplace
- 6:03where it's a little bit easier to query them, dig through them,
- 6:07see what's going on within your power automate environment. Yeah.
- 6:11So nifty a little bit
- 6:16maybe limiting in its first iteration, but that's okay. It's in preview.
- 6:21I should have called that out just to reiterate like preview not for production,
- 6:25blah blah blah. So maybe it gets better.
- 6:28So there's probably some things for folks to watch out here.
- 6:33Four if they've never done done App insights before.
- 6:38So App Insights is like you mentioned,
- 6:41it's its own little service that sits on its side and takes a pretty heavy
- 6:46dependency on yield log analytics and log analytics
- 6:51workspaces.
- 6:51I think Aveta like a wrapper layer on top of log analytics that gives you
- 6:57ultimately a way to interact with the data in the background in your app.
- 7:00Insights in the same way that you would with the log analytics workspace.
- 7:03It's our old friend Custo, again sitting there ready to go. But I,
- 7:07I think a couple of cool things that happen along the way here,
- 7:11like even with the limitations, like you mentioned,
- 7:14there's only a couple different types of telemetry you can catch.
- 7:17It's not like you can put like a custom event in your power app that's then
- 7:21writing to that login analytics workspace.
- 7:22Unless you wanna wire that all up yourself. There's no built in nicety for that.
- 7:28But once it's all there,
- 7:30now you've got all of the requests for your cloud flow
- 7:35runs and then all of your cloud flow triggers and cloud flow actions that you
- 7:39mentioned. Uh, they also land in a dedicated table as well for dependencies.
- 7:44So once all this stuff is sitting there and it's ready to go,
- 7:49then you can start getting nifty with it because there's other things that you
- 7:54can start to do.
- 7:55What if you wanted to create a an an alert
- 8:00through App Insights and Azure Monitor that goes and
- 8:05looks at a particular flow and actually alerts any time it fails,
- 8:10hey now you can just spin that up and do it.
- 8:13So that's all nicety and and ready to go.
- 8:15You mentioned multiple environments.
- 8:19There is a note in the documentation,
- 8:21not in the blog post shockingly that calls out when this
- 8:26feature goes from preview to ga,
- 8:29it is only going to be available for managed environments.
- 8:32So you're going to have premium usage rights for
- 8:37users on those power apps and I don't know why
- 8:42they're allowing you to do it in without premium use rights and
- 8:46without managed environments in the preview. That seems.
- 8:50Yeah, that's odd. It's like a bait and switch ish.
- 8:53A little bit.
- 8:54I call it out because some folks don't always go and read the documentation
- 8:59before they spin this stuff up.
- 9:01Yeah, like me, were you referring to me there Scott . But yes,
- 9:06I agree to give it to.
- 9:07I was ringing to us there to.
- 9:11To give it to everybody and then pull it away. That one is a little odd.
- 9:13We should probably go talk about premium or managed environments at some point
- 9:17in time too. I don't know that we've talked about those much add.
- 9:19Add it to your list. Yeah.
- 9:21Your other thing about K Q L too is nice 'cause I also have some flows that fail
- 9:25once because of illegitimate failure. Uh,
- 9:29so whatever triggered it didn't get passed in, it failed.
- 9:33So maybe even my alerts for a single failure are different versus maybe I
- 9:37wanna go see if this fails like 10 times in a row or fails 10 times in a
- 9:42certain time period.
- 9:44I want to be alerted of multiple sequential
- 9:49failures or multiple failure failures in a certain timeframe that maybe indicate
- 9:54it wasn't something with just that whatever happened to trigger that flow that
- 9:58particular time,
- 9:59but something more indicative of a more significant issue with
- 10:04my flow.
- 10:04Like a service account getting disabled that now causes all
- 10:09my connections to stop working or something like that.
- 10:12Never happens.
- 10:13That no never.
- 10:15Or it was tied to a user instead of a service account and said user left
- 10:20the company and their account get disabled. That never happens either. .
- 10:26Yeah, all those little things.
- 10:28We might have to come back and revisit this one once it GA's and see how things
- 10:33change with it along the way.
- 10:36But I'm a big fan of App Insights so I would take having this data in there
- 10:41over not having it in there.
- 10:42The other thing that you have to watch out for when you spin these up is
- 10:47it can create an app Insights resource for you or you can go down the
- 10:52path of saying I already have one. So in your case,
- 10:54maybe you have a couple apps and they're even across environments or things like
- 10:58that and you wanna tie 'em all together in the same app, insights, workspace,
- 11:01you can do all that.
- 11:03You just have to be mindful in the setup that you're going down that path and
- 11:06not letting Power Automate make some decisions for you. Ooh,
- 11:10app insights for everybody kind of thing.
- 11:12Absolutely. Another.
- 11:13Governance thing to watch out for.
- 11:15Yes, there's a lot of governance stuff with the power platform as a whole.
- 11:19It's a,
- 11:20it's a unique service within the platform from a governance perspective.
- 11:25Platform is a unique service. There you go. Good way to sum it up.
- 11:29What's next Scott? Which one are you gonna jump to? What.
- 11:32I'm gonna keep us in? Platforms. Platforms. Did you see,
- 11:35I know you're always excited about this stuff as as our Mac OSS reporter
- 11:41platform,
- 11:42SS s o for Mac OSS is coming
- 11:47to
- 11:49intra ID tenants near you as a deployable,
- 11:53extensible agent on Mac OS clients.
- 11:58I think this one's kind of cool.
- 11:59This one is cool and you're right.
- 12:01I am excited for this one because up until now we did have
- 12:07the Microsoft Enterprise S SS O plugin for Apple devices.
- 12:11So this essentially allowed you to set up S ss o
- 12:17once you were logged in so that you could SS s o into various applications
- 12:22on your Mac OSS device.
- 12:24Now it is coming to Mac OSS as a whole.
- 12:27So this is taking that SS s O plugin bringing it up to the next
- 12:32level where now you can have your platform credentials for Mac oss Go
- 12:37Passwordless using Touch id unlocking the devices and be
- 12:42signed into intra ID under the hood.
- 12:45So essentially now you go log into your macros device and you're signing
- 12:50into intra ID as well using and they go into all the,
- 12:54it's the device bound,
- 12:55cryptographic key phishing resistant credentials based on
- 13:00the windows Hello tech and some of the Apple hardware that's already there.
- 13:04So this is gonna be cool.
- 13:06It's not quite Azure AD joining your device as a whole,
- 13:10but it's bringing it closer to that where now you're
- 13:15again signing into your device, signing into enter ID at the same time.
- 13:19So this is enabling all the SS s o for all your applications.
- 13:23You can also with this now synchronize your local account password and the
- 13:28enterra ID password.
- 13:29So your user account password on Mac OSS is also your enterra ID
- 13:34password.
- 13:35It doesn't sound like it's necessarily the same account or even the same user
- 13:40id,
- 13:41but at least the passwords on the backend are going to be the
- 13:46same.
- 13:46So you don't have to remember two passwords if you even use a password for
- 13:51anything anymore. Most of mine is all touch, ID watch, ID passwordless,
- 13:56all of that type of stuff.
- 13:58This is an upcoming public preview of platform SS s O for Mac
- 14:03oss and it'll work with Intune.
- 14:06It is coming to other M D M providers soon.
- 14:09I would say the timing for this is interesting.
- 14:12So it does say your Mac OSS requirements, I saw it in here somewhere,
- 14:17venture and hire,
- 14:18but it was also announced the same day that Apple announced their September
- 14:23event for Mac,
- 14:25which is usually also when the next version of MAC OSS comes out.
- 14:29So it'll be interesting to see.
- 14:30At first I thought maybe this was going to be a Sonoma specific functionality
- 14:35that Apple did something, but with the requirement of enterra,
- 14:38maybe there's some updates coming to ventra as well or just
- 14:43Microsoft finally was able to incorporate this into enterra a nice improvement
- 14:48for MAC users that are using intra id
- 14:53formerly known as Azure ad. Yeah.
- 14:56So the way I ran into this one, uh,
- 14:58you probably ran into it through the R S SS feeds that that we kinda automate
- 15:03and bring in. I ran into this one over on,
- 15:07I think it was Mastodon.
- 15:08Somebody was chatting about it and it it was one of the PMs in
- 15:14Azure AD and the way he framed it is like, hey,
- 15:19we're finally doing this. You can finally bring your C T o,
- 15:23your security leads and your devs on board . I was like,
- 15:28that is so true. Like I,
- 15:31I remember I I worked for for a logistics company here in Jacksonville and
- 15:36our C E O was the only user in the organization with a Mac . That was it.
- 15:41He liked a Mac the rest of us like we bought Lenovo's stuff like it
- 15:46was free candy but the C E O wanted his Mac and that
- 15:51was like the pain point for all of us to live with.
- 15:53And then I was thinking about it where I was like,
- 15:55oh and our chief security officer used to be on a Mac too.
- 15:59Yeah that that made life more difficult back then like it's so true.
- 16:03Bring folks into the fold and get 'em where they need to be.
- 16:06I'd be interested in seeing how this one works
- 16:11there.
- 16:12There's other components of Intune and the whole M D M stack on
- 16:17a Mac that maybe don't integrate the nicest.
- 16:22So one area for me would be say you impose like FileVault restrictions,
- 16:26which is basically like encryption on the MAC side. Yep.
- 16:29You might have BitLocker on Windows kinds of things.
- 16:32Does that extend all the way down to FileVault?
- 16:35Do you still need a local password for FileVault? What does that look like?
- 16:38How does it come through?
- 16:40'cause when you do things like BitLock or like you get your recovery keys stored
- 16:43Azure ad your help desk can work with you.
- 16:46Some of those constructs don't necessarily exist.
- 16:48Like they're not the same on Mac OSS Ventura to all the way
- 16:53up to the upcoming Sonoma. We shall, we shall see. Yeah.
- 16:58It's.
- 16:58Always interesting when stuff like this comes out because I think
- 17:03sometimes from a marketing public perception side of things,
- 17:08it's like Microsoft and Apple are bid rivals and they can't stand each other in
- 17:13all of this. But then you see stuff like this and it's, you wonder how much,
- 17:18and I suspect there's a lot more collaboration between the two companies,
- 17:22especially when it comes to stuff like this or I know even some of the stuff
- 17:27with OneDrive files on demand.
- 17:30I think one of the sessions I saw there was some partnership in working between
- 17:33'EM but from getting their systems to work together.
- 17:37When it comes to things like the SS s o when certain features that Microsoft
- 17:41maybe rolls out,
- 17:42I get the feeling maybe there's a little bit more partnership together behind
- 17:46the scenes than maybe public perception always gives to .
- 17:51Give you some of this stuff 'cause it is important. Like you said,
- 17:53you want your C-level guys to guys girls to be
- 17:59brought into the fold and your devs and your security people.
- 18:03Oh you also forgot Scott. Marketing Graphics,
- 18:08multimedia. Those people are also usually all on their Macs.
- 18:11They are indeed right here.
- 18:15Do you feel overwhelmed by trying to manage your Office 365 environment?
- 18:19Are you facing unexpected issues that disrupt your company's productivity?
- 18:23Intelligent is here to help much like you take your car to the mechanic that has
- 18:27specialized knowledge on how to best keep your car running Intelligent helps you
- 18:31with your Microsoft cloud environment because that's their expertise.
- 18:35Intelligent keeps up with the latest updates on the Microsoft cloud to help keep
- 18:39your business running smoothly and ahead of the curve.
- 18:41Whether you are a small organization with just a few users up to an organization
- 18:45of several thousand employees they want to partner with you to implement and
- 18:50administer your Microsoft Cloud technology,
- 18:53visit them at intelligent.com/podcast.
- 18:57That's I N T E L L I G I
- 19:01k.com/podcast for more information or to schedule a 30
- 19:06minute call to get started with them today.
- 19:09Remember intelligent focuses on the Microsoft cloud so you can focus on your
- 19:13business.
- 19:15Alright, so yeah platform SS s O for Mac,
- 19:20OSS nifty stuff.
- 19:22Why don't we keep talking about intra ID since
- 19:27we're over here.
- 19:28There's another interesting announcement that came out this
- 19:33week-ish. Uh this week.
- 19:35Yeah.
- 19:36Any, anyway,
- 19:37it'll have come out by some week by the time somebody listens to this.
- 19:40But that is that there is the ability now
- 19:45to do a p I driven provisioning within your
- 19:50intra ID tenant.
- 19:52So rather than having maybe some outside service that does some wonky
- 19:58synchronization or something else that comes in along the way,
- 20:02you can just go ahead and actually take your system of record.
- 20:08I don't know,
- 20:08let's say I've got like SS A P or SuccessFactors or something like that.
- 20:13I can take the data out of there and as long as I can extract it and get it out,
- 20:18you can then push those changes like when they occur in that system of record
- 20:23over to this new A P i which sits within the
- 20:28uh,
- 20:28entra provisioning service and then that goes ahead and pushes that
- 20:33data out to where it needs to be.
- 20:37And the cool thing is it's not only like cloud provisioning 'cause now you've
- 20:41got,
- 20:42because we we've got the whole like intro cloud sync thing and and you
- 20:46do have hybrid like you mentioned putting all those things together.
- 20:50So now you can push into a single A p I endpoint and then say
- 20:55it's something that needs to affect change in your on-premises
- 20:59active directory. Great.
- 21:01That change can flow straight down from the provisioning a P I and the
- 21:06provisioning service down into what's yet another local agent running
- 21:12behind your firewall and ready to go in the form of the provisioning agent.
- 21:16And that can do things like push users and push changes into your on-premises
- 21:20active directory and then if you have synchronization configured between that
- 21:25and your intra ID tenant cloud sync kicks in, connect sync,
- 21:29click whatever you happen to be doing a along the way to get you there.
- 21:35So I think that's a super powerful one.
- 21:38Kind of seemed to go out under the radar.
- 21:41I didn't see a bunch of fanfare about it but I think it's super powerful,
- 21:46super cool and a long time coming. Yes.
- 21:50I agree and I have some customers that this might be beneficial for.
- 21:55Like I've done some weird stuff, I'm not gonna lie,
- 21:57it gets a little hacky at times but like one customer.
- 22:00That's janky, janky.
- 22:02Janky, it's not hacky, it's janky. What about convoluted?
- 22:05Convoluted is a good one too.
- 22:06I've done things for customers like their system again,
- 22:09they have a different system of record that they need everything to go into
- 22:13Azure AD four and it's things like C SS V files getting emailed
- 22:18out of this system because they can act do like regular export
- 22:23email reports and then we pull the said C S V out of the
- 22:28email attachment, dump it into blob storage,
- 22:31which triggers another flow that runs a runbook to go parse the C SS V,
- 22:36find all the changes and synchronize them up to Azure Active directory.
- 22:41Just being able to tie straight into an A P I if you can go straight from that
- 22:44system of record instead of all the secret sauce in between them,
- 22:48this would be very nice And I was talking to another customer that was in the
- 22:51same type of boat, different system of record,
- 22:53wanted to just shove everything straight into N I D and
- 22:59this is going to be a good solution.
- 23:01So this is one that is absolutely on my list and I'm excited to play about play
- 23:05with because I do have some customers in this scenario where we might
- 23:10be able to simplify some things for them.
- 23:12One of the other things to call out in here is
- 23:17you can give it a try and the a p i driven provisioning,
- 23:22this is another one of those preview features. Don't use it in production,
- 23:25yada yada yada. You can start using it if you have Microsoft enterra ID P one.
- 23:30So premium P one license for Azure ad enterra but then it also
- 23:35says licensing's terms will be released at general availability.
- 23:39So if you are a brave soul that wants to try this in preview,
- 23:44it doesn't sound like they are necessarily committed to this
- 23:48only being or remaining just a P one feature
- 23:54that this similar to the log analytics one we
- 23:59talked about or the app insights could change once
- 24:04it hits GA from that licensing perspective and what's required. Again,
- 24:09I know they want people to try it, right?
- 24:10That would be my guess as to why they're doing this is we have this license now
- 24:15but we don't know what it's gonna be when it comes out in general availability.
- 24:19And they've done that before too with other stuff or it's, it's free now.
- 24:23It's not gonna be free once it comes out.
- 24:25I wish they would just release the licensing terms right away so that people
- 24:29would know is this even worth playing with because
- 24:34I could play with it but there's no way I'm gonna be able to upgrade everybody
- 24:38to premium P two or pay an extra $5 per user a month
- 24:43to have a p I driven provisioning because it's just not feasible for their
- 24:47budget. So I don't know.
- 24:49It'd be nice to see what the licensing terms are gonna be once it hits general
- 24:53availability. Even if you have different ones in preview. In preview,
- 24:57you can use it here once it, it's a general availability, it's going to be this.
- 25:01It just helps from a planning perspective in my opinion. Yeah.
- 25:04So there's some other interesting things I think about the provisioning service
- 25:09as a service kind of thing and the way it's been implemented.
- 25:12So lots of the identity management world,
- 25:16like if you think about common, so directory tooling,
- 25:21so stuff like, I know SailPoint does this,
- 25:25I think Okta does this well off the top of my head but there there's a bunch of
- 25:28them that kind of interoperate based on a standard called Skimm.
- 25:33So the system for cross domain identity management
- 25:38I I believe that is.
- 25:39But Skimm is basically it's an open standard
- 25:44and it allows you to automate user provisioning end to end
- 25:49across all these various systems that also implements Skimm.
- 25:53One of the things that comes along with Skimm is bound schemas
- 25:58to work with in different scenarios and the a p i provisioning service
- 26:03supports both of those.
- 26:05So it gives you the ability to map from the Skimm core
- 26:10user schema and the enterprise user extension all the way back to the
- 26:14attributes in Azure ad.
- 26:17I think an interesting thing here is even though the payload to talk
- 26:22to the A P I provisioning service is skimm content,
- 26:26like it's actually a specific content type that gets fired into the A P I.
- 26:32So like the content type header on these requests gets set to application slash
- 26:36skim plus json it. It is not a Skimm compliant endpoint.
- 26:41So even though skim is the glue that's potentially holding things together,
- 26:45Microsoft is not fully coming into that ecosystem.
- 26:49So it's a one-way inbound endpoint. This is inbound provisioning,
- 26:54it's not outbound provisioning and full right back to these systems
- 26:59as well. So I think that's important to call out and you do have to
- 27:04consume a custom a p I here it's not,
- 27:07hey let's go ahead and just fire this up into existing product
- 27:12stacks that natively integrate with Skimm.
- 27:14Like you've gotta do some work to get it going.
- 27:18It is not a standard Skimm endpoint by any
- 27:22definition of the record.
- 27:24It's also very different from I think graph.
- 27:28If folks are looking at it and they're saying like,
- 27:30oh why wouldn't you use the graph to do this?
- 27:33Fundamentally different things like just the way the APIs are built out and what
- 27:37they're made to do. Like graph as being very o data centric, single user,
- 27:43not very bulky. This is all just bulk upload,
- 27:47bulk provisioning, get it into this schema, get it into this format,
- 27:50do some mapping for us and we'll make the rest of the magic
- 27:55happen in the in the background.
- 27:59Which I think it's very good to see it.
- 28:02I wonder how that changes over time too though if it does need to be
- 28:07become something that plays nicer in the ecosystem to get folks to latch onto it
- 28:11or if they just go, oh my gosh,
- 28:12I've been waiting for this so long that I'm going to take it and run with it
- 28:17kind of thing.
- 28:17Yeah, they have a big explanation of why it's not a standardized skim endpoint.
- 28:22There's a lot of good information about this in the
- 28:27F A Q that we threw it in the discord chat,
- 28:29but if you're listening to this after the fact,
- 28:31it is also will be in all the show notes.
- 28:35If you are thinking about implementing this,
- 28:38definitely go read through these frequently asked questions about the
- 28:43a p I driven inbound provisioning because it's not just a,
- 28:47there's a lot of information we'll leave it at that.
- 28:49There's a bunch of information in this few frequently asked questions about more
- 28:53details on how it's all working, why it's all working the way it does,
- 28:58all of that type of stuff.
- 28:59Indeed. All right, what else? We we we have time for one more, for one more.
- 29:03You wanna talk about, you wanna talk about Python? We.
- 29:06Can talk about Python.
- 29:07You were doing such a good job with your flow from one to the next and then we
- 29:12just completely band all that and said let's talk about Python ,
- 29:15but let's talk about Python. I.
- 29:17Ran outta things so , this'll be our, this'll be our last one. There.
- 29:22Was my G P O one. We can do that one next time though. We'll do Python. We'll.
- 29:25Save that one in the list.
- 29:27So Python used all loaf for the place in
- 29:32data science land.
- 29:34It's available in your favorite products of choice.
- 29:38Like maybe you've spun up Databricks and you're like Ooh,
- 29:42I'm gonna run a notebook over here and get things going and start playing with
- 29:45some pandas data frames and things like that.
- 29:48If you've ever done that on that side and you're like, huh,
- 29:51I really wish I could do this in Excel now you can do
- 29:56pandas data frames and pandas extensions and all the things that come along
- 30:01with that ecosystem within Excel except you're
- 30:06not really doing it in Excel.
- 30:08And I think this is cool the way they set this one up and get it ready to go.
- 30:13So if you hop over to the latest insider
- 30:18builds of office within Excel in public preview,
- 30:23now there is Python functionality.
- 30:27So rather than having a standard Excel function,
- 30:30you can come over and you can run Python functions
- 30:36which are just using pandas,
- 30:38which is super cool because there's tons of plugins and other things for like
- 30:43charting and visualization and all that stuff or running specific types of
- 30:47algorithms on top of your data. Like all that stuff exists in conduct.
- 30:51Super easy to pull in.
- 30:52Like that's all supported over here.
- 30:56So if you're a data analyst and you've been pining to have
- 31:01access to data manipulation and working with data frames inside
- 31:06of Excel, I think this is a super cool thing.
- 31:11Just like not only on the data manipulation but also the visualization front.
- 31:16There's all sorts of other things that go into predictive analytics,
- 31:21modeling, statistics modeling, forecasting,
- 31:24all those kinds of things that now you can also
- 31:29go ahead and
- 31:32run through this engine and if you're already in the ecosystem,
- 31:36like it should be super easy to come over like it's all pandas,
- 31:41it's all Conda, Anaconda and that stack.
- 31:45It should all just work, which is cool.
- 31:49The even cooler thing I think is how it all just works,
- 31:55huh?
- 31:55Like how did they do all this and how did they bring this functionality down to
- 32:00a local, a local Excel client and make it consistent?
- 32:06And the interesting thing is they didn't bring Python down to the
- 32:11client.
- 32:12They brought the data from the client up to the cloud
- 32:17specifically in the form of a container.
- 32:20So Microsoft is hosting Azure container instances that
- 32:25are using pre-built packages from Anaconda
- 32:30that then you can go ahead like I said and and absolutely bring in and inject
- 32:34your own packages into that container.
- 32:36But effectively Microsoft is running like compute as a service in the cloud in
- 32:40the form of these Python containers that are all stood up and ready to go that
- 32:46your Excel client can talk to.
- 32:49So it can pass data to that container that gets sent out
- 32:54via this specific Python function that's now available in Excel.
- 32:58Like it's the Excel function. So Excel,
- 33:01open paren and then close paren and it can just
- 33:06return that data in the form of a data frame that can then talk locally
- 33:11inside of Excel.
- 33:13I think all sorts of interesting kind of things to follow
- 33:18up on here, but I'm excited about the Python part of it.
- 33:21Anytime I don't have to go and and open a notebook and hop over to a specific
- 33:26workspace or anything like that is really nice.
- 33:30I wonder how much customers will latch onto it once they realize that the
- 33:35units of compute that are running things are actually container instances
- 33:40and sitting out there in a hosted on behalf of managed
- 33:45on behalf of environment within Microsoft land.
- 33:49There is a little bit of an article out about this
- 33:54functionality, I'll put a link in the show notes.
- 33:56It's called Data Security and Python in Excel but it talks specifically
- 34:01about what these Azure container instances are,
- 34:06where code runs and what your Python code has
- 34:11access to and does not have access to.
- 34:15So I think it's important to go and read through that because
- 34:20effectively like now your Excel client is punching out to random place
- 34:25on the internet to go ahead and put data in and out of your Excel notebooks.
- 34:29Yeah it is interesting that it's doing it all in Azure. So I have a question.
- 34:33I've not done a bunch with Python, I don't do much data science,
- 34:38this is not on my list to go play with but just from
- 34:42understanding it,
- 34:43is it that the amount of data in the amount of compute that Python
- 34:49would use to crunch this is that it needs something like a container in Azure?
- 34:53Like why build this to run an an A C I instead of
- 34:58building this to run locally?
- 35:00Or is it just that there would be so many more requirements on how you'd have to
- 35:04set it up locally because, and this came out, I actually had a,
- 35:09a friend of mine that asked this question because in preview it's gonna be
- 35:13included but this is another one that after preview some fun,
- 35:16some functionality they don't specify what will be restricted without a paid
- 35:20license. And he was like,
- 35:21I don't wanna go learn all of this and set all this up but while it's in preview
- 35:25only to lose this functionality because I'm not gonna get my company to go buy
- 35:30a paid license for me to do it once it comes out of preview.
- 35:34Why isn't there some way that I can still use Python and Excel but run it like
- 35:39using my local compute instead of having this dependency on
- 35:44a C i or maybe you can already do this all locally if you go set everything up.
- 35:48That was just one of those I was curious about not having done a ton with it.
- 35:51I didn't necessarily have a good answer. I think.
- 35:54Part of it is the ephemeral nature of
- 36:00data science.
- 36:01So if you think about these massive lakehouse and data lakes
- 36:06and other things that kind of sit out there today, right? Uh,
- 36:09there might be some form of like centralized governance that sits on top of
- 36:13that.
- 36:13But quite often you have like disparate communities that are actually working
- 36:17with the data.
- 36:17Like you might have one set of data scientists working with say say do the
- 36:22medallion architecture, right? You're running everything through like bronze,
- 36:25like raw data, refining the silver, getting it out to gold, blah blah blah.
- 36:28Like all that data is meant for different audiences and for different
- 36:33types of analysis.
- 36:36So quite often while you maintain like the centralized data source,
- 36:40the lakehouse,
- 36:41now let's imagine like your Excel file is the centralized data source. Yep.
- 36:45The people who consume that data are different audiences and they're gonna go
- 36:48about it in different ways with potentially different tools.
- 36:52So you might spin up like like a Databricks cluster to do some
- 36:57one-off analysis on an already refined dataset 'cause you wanna go and look at
- 37:02it in a different way or there's a different like statistical model you wanna
- 37:05run against it, something like that. So you'll spin that up, you'll create it,
- 37:09you'll do it all, you'll crunch all the numbers,
- 37:11you'll run through it all and then you'll spin it down 'cause you got the result
- 37:15kind of thing and and I think that ethos of ephemeral
- 37:20analysis that potentially doesn't always need to stick around plus the
- 37:26ephemeral nature of maybe you are doing one type of analysis and I'm doing it
- 37:31another and we're gonna approach that in different ways with a different set of
- 37:34plugins.
- 37:34All those kinds of different things actually makes a ton of sense with
- 37:39something like container instances which are also quite ephemeral in their
- 37:44nature, right? Spin 'em up, spin 'em down,
- 37:47only use the compute when you need it and then hey the rest of the time I don't
- 37:51need the compute, let's just ditch it out the side and get rid of it.
- 37:56So I imagine like that's a big component of it is the very nature of some of
- 38:00this stuff is ephemeral works end to end if you think about it that way.
- 38:04The other one is consistency and availability of these things.
- 38:08Like containers make a great way to spin up
- 38:13packaged environments that are the same for everybody at the beginning and then
- 38:17they can deviate as needed but whenever they deviate too far like they can just
- 38:22reset and come back to baseline and start again. That kind of thing.
- 38:26I think it all comes together and works.
- 38:29It's an interesting way to solve the problem and I think it
- 38:33opens up the ecosystem to a bunch of new folks.
- 38:37So if you're thinking about it from the context of oh my gosh my company is not
- 38:42gonna pay for a license for me to do this in Excel,
- 38:44is your company paying for you to have access to Databricks or
- 38:50Synapse or Snowflake or like one of these other.
- 38:53One of those other ones.
- 38:54Analytics platforms? 'cause if they are,
- 38:58I'm gonna bet that the cost bakes out to be a lot less to run a container
- 39:02instance to run on top of your Excel workbook than it is to spin up an entire
- 39:07unit of compute around even like a single node Databricks cluster.
- 39:10Got it. That makes sense. And I guess the other aspect of it,
- 39:13I was going through the article here too is they do talk about that it's built
- 39:17for teams so you can also like store these workbooks in
- 39:23SharePoint. You can put them in Microsoft Teams,
- 39:25which is still putting them in SharePoint, send them via Outlook,
- 39:29which you shouldn't be emailing files via Outlook anyways,
- 39:32but it still has all the comments,
- 39:33the mentions and then as different people maybe go grab these notebooks,
- 39:38to your point,
- 39:39your Excel workbook is your source of data and maybe different people have
- 39:43different worksheets,
- 39:44all of that or other people are working with you on the data,
- 39:48they'll be able to open it and not all of a sudden run into,
- 39:51my machine isn't set up the same way your machine is from a Python perspective
- 39:55so it's not working. You just open the workbook and it sounds like it's there.
- 40:00I'm guessing it just as soon as someone opens it,
- 40:02it maybe spins up a new a c I instants in the background to be able to do
- 40:07anything with it.
- 40:08They say sensitivity labels work with all of this for protection policies but
- 40:13this also probably better facilitates the portability
- 40:18of an Excel workbook. I guess they would be curious,
- 40:21does it only call out when it re renders data?
- 40:23Let's say you have the license for this,
- 40:25you go do a bunch of data analytics around the podcast in Excel,
- 40:29you send it over to me or share it with me and I don't have a license,
- 40:34is it just not gonna work for me?
- 40:36It'll be interesting to see how that works too if you share these with people
- 40:40that are licensed differently for Python and Excel.
- 40:45T B D, we'll see, I haven't seen that one yet.
- 40:49There's a really good video out there. There's this awesome person out there,
- 40:54Layla Ani, she's an office M V P like she's great. She's got a YouTube channel.
- 40:59If you go in you watch like John Savile videos for Azure stuff like just go
- 41:03watch her videos for office stuff.
- 41:06Ooh Excel word one like stuff like I'm not excited about all the time but
- 41:11she has an awesome video that walks through this
- 41:16feature I think in a really great way.
- 41:20Like it takes you through like the baby steps of okay,
- 41:23let's bring in our first data frame. Let's see how a data frame plays over here.
- 41:27Talks a little bit about what data frames and what they are and let's leave the
- 41:31whole a c I thing and like the technology and the background out of it.
- 41:33Like just here's what Python and Excel potentially means for you with a nice
- 41:38simple dataset.
- 41:40I think rather than folks like getting on their jump to conclusions Matt and
- 41:43trying to worry about licensing and how it all works in the background.
- 41:47Like maybe just go see if the functionality works for you and the functionality
- 41:51works for you and the values there.
- 41:53It's a lot easier to have that conversation about like downstream,
- 41:56R O I on on licensing and all those kinds of things because if you get like
- 42:01analysis paralysis in the beginning, I think for something like this,
- 42:04like you're potentially missing out what could be a really cool tool in your
- 42:09toolbox.
- 42:09Right?
- 42:10I wonder if it'll end up being even like a pay as you go type license or you can
- 42:15buy like credits thinking back to the whole spin up a container as units
- 42:20of compute.
- 42:20We've seen more of this pay as you go type stuff coming in through some of the
- 42:25syntax stuff that we talked about earlier too.
- 42:27Like I could see this one even being a pay as you go type model
- 42:32based on usage of those instances. So we shall see,
- 42:37but I do.
- 42:38We'll see it is compute somewhere, right?
- 42:40Like let's be clear like it is compute that's running all the other compute
- 42:44things.
- 42:44Like you look at like dev boxes and cloud PCs and all this other stuff that runs
- 42:49in the background in Microsoft land.
- 42:50Like part of the reason that stuff costs money is because it costs Microsoft
- 42:55money to run it.
- 42:55Like you're literally firing up a server for some period of time.
- 42:59So keep that in mind too about licensing models and things like that.
- 43:02Sometimes underlying technical implementation has to drive some of that just to
- 43:06get those wonderful cogs back.
- 43:08Yep. But I do like to that point of the portability,
- 43:11it definitely is a better way to do this I think from a portability standpoint.
- 43:15So who knows what it will end up being,
- 43:17but definitely some cool features and functionality coming to
- 43:22Excel. And with that we should call it a day.
- 43:25The leaves are still gently waving outside my window with an occasional wild
- 43:30gust or two,
- 43:30but I also have clients pinging me that they have an urgent issue that I
- 43:35should probably go attend to. It.
- 43:37Happens. All right, cool. Go fire up Spotify.
- 43:40Start playing Rocky like a hurricane,
- 43:42like a little bit of scorpions in the morning to get you going and you'll be all
- 43:46set.
- 43:46Should I play it loud enough that I can go stand in the middle of the street
- 43:49with my flag and brave the hurricane?
- 43:53Be the Florida man . It was funny. One last story.
- 43:56I had a friend that totally did this the other day.
- 43:58I was on a conference call with him, he's down in Tampa and he was like,
- 44:02oh this is a good, a good squall coming in. I'm going outside.
- 44:06And he literally like left in the middle of the conference call to go run
- 44:09outside in the middle of the squall and he didn't come back for 20
- 44:13minutes. I was like, see outside, like hanging onto a stop sign,
- 44:16blowing sideways or, but I was like, this is, it's definitely Florida.
- 44:21Yeah.
- 44:21Yeah, definitely Florida. That's our new motto now.
- 44:25Definitely. Definitely Florida . Alright, thanks Scott.
- 44:28Enjoy the rest of your day.
- 44:29Hopefully you are able to maintain power throughout the day today and we will
- 44:34talk to you later. All right, thanks Ben. Yep.
- 44:38If you enjoyed the podcast, go leave us a five star rating in iTunes.
- 44:42It helps to get the word out so more it pros can learn about Office 365 and
- 44:47Azure.
- 44:48If you have any questions you want us to address on the show or feedback about
- 44:52the show, feel free to reach out via our website, Twitter or Facebook.
- 44:56Thanks again for listening and have a great day.