Episode 65: Cloud Coreyography Mark 2 with Azure’s Corey Sanders

Episode Summary

Another chat with another Corey but this time Corey Sanders has a shiny new title: Corporate VP of Microsoft Solutions. In this episode the two Coreys discuss the Microsoft mission statement, more real-world Azure examples, and why “vendor lock in” is both simpler yet more complicated today than ever.

Episode Show Notes & Transcript

About Corey Sanders

Corey Sanders has 15 years of experience at Microsoft with 13 years of managerial experience. In the last 9 years, Corey has been in the Azure team building the Azure Compute service, and he recently moved into a new role as Corporate Vice President for Microsoft Solutions.

Links Referenced


Announcer: Hello, and welcome to Screaming in the Cloud, with your host, Cloud Economist, Corey Quinn. This weekly show features conversations with people doing interesting work in the world of cloud, thoughtful commentary on the state of the technical world, and ridiculous titles for which Corey refuses to apologize. This is Screaming in the Cloud. 

Corey Quinn: This episode of Screaming in the Cloud is sponsored by N2WS. There are a number of backup solutions that are available in AWS including the recently announced AWS Backup. Well, AWS, back the #$*% up. Backups are incredibly easy. Restores, however are absolutely not. You want to find out whether your backups worked well in advance instead of the way that most of us do: when they don’t work quite right immediately after you really, really, really needed them to work correctly. Check them out at n2ws.com. That’s n2ws.com. Thanks to them for supporting this ridiculous podcast.  

Corey Quinn: Welcome to Screaming in the Cloud. I'm Corey Quinn. I'm joined, for a second time, by Corey Sanders.

Corey Sanders: That's right. I think the second time is supposed to be better, but we'll see. No promises.

Corey Quinn: So they tell us. The last time that we chatted a year ago, here as well at Build, you were the Corporate VP of Azure Compute.

Corey Sanders: That's right.

Corey Quinn: You're now the Corporate VP of Microsoft Solutions. What is that?

Corey Sanders: You know what, when I figure it out, I'll let you know. No, just kidding. Well, one important aspect is I moved from the Engineering Product Team into the Sales Team, into the Technical Sales Team, so that's a pretty big shift. I'm responsible for the four big solutionaries that we sell as a company. That includes data and AI, apps and infrastructure, which together become what we know as Azure. But then also Modern Workplace, which is Office 365 and Windows Client and then also business applications, which is our dynamics product. I'm sort of responsible for all of those. That's the new gig.

Corey Quinn: It sounds like it's a lot of work, to be very blunt.

Corey Sanders: It's a lot more work than I want, let's be honest.

Corey Quinn: You're doing Microsoft solutions, other people are causing Microsoft problems.

Corey Sanders: That's right.

Corey Quinn: It winds up sort of this wonderful balance.

Corey Sanders: That's right. I get called in ... I'm sort of the fireman, as it were, but without any running water, so we'll see what happens.

Corey Quinn: This morning there was the Imagine Cup World Championship, where you were on video from the expo hall, which looked like it was 30 feet away from where the rest of us were sitting. One thing that you mentioned was that it turns out as the Microsoft mission statement, which I didn't realize companies still had, but you folks do. "To empower every person and every organization on the planet to achieve more." Now, normally, I tune out on those things, but you were wearing a t-shirt in the video, which is a bit of enough of a departure from what everyone else was wearing. "Okay, I'm going to pay attention to this guy."

Corey Sanders: "He must know what he's talking about."

Corey Quinn: Exactly. And if he didn't know, let him dress like this. I'm giving him the camera.

Corey Sanders: That's right. Yeah, well, the wearing of the t-shirt ... I'll answer that question first and then I'll dig in because I think ...

Corey Quinn: Start with the easy one.

Corey Sanders: Oh, man, yeah. I mean the wearing of the t-shirt, this has become sort of, in some ways, my brand, which is a little bit of a weird thing that you'd day your brand is that you wear t-shirts. But as I moved over from engineering to the sales organization, it was sort of like I continued to just wear t-shirts. And then it became sort of a thing that, "What funny, weird t-shirt is this guy going to wear?" That now I have to wear t-shirts. I don't actually even have an option anymore, because everyone's like, "You're wearing a dress coat. What are you doing, man? You look like an idiot." And so t-shirt all the time.

Corey Quinn: "Are you here to fire someone? What's the story? Oh, you're interviewing for another job? What's going on?"

Corey Sanders: Yeah, exactly, "I hope the interview went well." That's the thing with my t-shirts. I have to continue to buy cool t-shirts. So if you've got any ideas, let me know.

Corey Quinn: Well, today was a plain gray shirt with no logo on it.

Corey Sanders: It was, which is like a marketing thing. I had logoed shirts, and they were all going to risk people suing us, so we ended up wearing blank.

Corey Quinn: Got you. So you fundamentally gave up on whole NASCAR approach to business models?

Corey Sanders: That's right. That's right, that's right. Apparently no one owns plain gray, so that's good. So that's what I wore.

Corey Quinn: Yeah, I'm sure someone out there is currently...

Corey Sanders: Someone out there's already filing suit. So then going back to your question about now that we finally got through that we can talk about the mission statement. I find the mission to be super inspiring. What an interesting mission statement to be so focused on what we can help others to do versus what we ourselves are doing.

Corey Quinn: Right. Other folks' mission statements tend to be somewhat inward looking. Like, "We're going to categorize the world's information and then sell it to people."

Corey Sanders: I have no idea who that is.

Corey Quinn: Or, "We are going to not rest until no one else can make money doing anything."

Corey Sanders: That's right.

Corey Quinn: And it's great. This is a little bit more aimed at helping other people achieve.

Corey Sanders: That's right. That's right. One of actually the more exciting parts about moving in this transition. When I was in the product team, I was out talking with customers a lot, figuring out what they needed, what they were looking to do, how our products could help them. And this shift to the sales organization has been pretty exciting because when you think of sort of a modern sales organization, it is not about, how do I get in there and try and get you to sign this paper and give us a check? It's all about, how am I helping you solve problems? How am I sitting down and understanding what issues you have and how we can bring the right resources from our side to partner with you.

There's been some really exciting examples of this. I think we may have talked about it a year ago with Walmart. Some of the work we did, we've got sort of the Joint Development Center with Walmart. We're doing really cool things with stores and an IoT-based solutions with tracking sort of the health of their refrigerators. There's been a lot of these really interesting opportunities for us to learn more about retail and them to take advantage of some of our deep technical understanding for the Azure services. When you think of the modern era of selling in the cloud, it really is around enabling and less selling, if that makes sense.

Corey Quinn: Which is interesting, just from a perspective of meeting customers where they are without feeling the need to be what your customer is. More or less, providing services that help empower them to continue doing the thing they're already doing without a lot of the toil goes into that.

Corey Sanders: That's right. Yeah. And in some cases even doing things better. Or doing things more interesting. I think, a good example also sort of retail is Kroger's, they've added as part of their shelves, they have this product called Edge and they add sort of an additional advertisement underneath the chips to help you sort of decide which chip you actually want to go by, which turns out as actually a really important problem that we have to go work through.

Corey Quinn: If they can solve that one for me, it saves at least 40 minutes every I go shopping.

Corey Sanders: Exactly. This is a partnership with them to sort of solve this experience, this selling-buying experience in a new way. Not only is it making existing problems solved in an easier way, but creating new solutions to problems that sort of they didn't even realize they had. It's been really exciting to be a part of that and to be sort of at the front line of those conversations from the sales side.

Corey Quinn: Talk to me a little bit about how you're viewing the, I guess, world of hybrid, where people start off on premises, generally speaking, there aren't too many born-in-the-cloud companies that have scaled out today on Azure that have been in Azure their entire existence and are now multi-billion dollar companies. Everyone has something legacy. There's always something that's vaguely greenfield. Last year, we spoke a little bit about Azure stack, 12 months later, how's that going?

Corey Sanders: It's actually very exciting. I mean, I think we've seen a huge amount of interest and growth on the Azure stack side. What we're seeing is that there's, not only a lot of opportunity for these hybrid deployments for customers who come in and say, "Great, I'm going to need both the cloud-based solutions, but also something deployed locally. And it's not just because I used to have something locally, but because there's something that requires me to stay local.

I'm running a manufacturing plant and I really can't depend upon the network to always be there. I'm going to have something local that will keep the manufacturing plant running, but then use the public cloud for additional data analytics, additional analysis and so on." That combination of sort of intelligent cloud, intelligent edge has become a really interesting cornerstone of our overall platform.

Corey Quinn: Which seems critically important. We all have a story from somewhere in our past, where we have a dependency built into something that's far away and remote, and invariably, the fiber line leading their encounters it's natural predator, the backhoe. Suddenly, the entire factory is down for want of a single fiber connection.

Corey Sanders: Right.

Corey Quinn: And we have these agonizing stories that take 24 or 48 hours to get resolved during which time nothing happens. And the story of, "Oh, everything should live in the cloud. It should just be okay," simply isn't tenable when you're talking about significant volumes and significant scale here in the real world.

Corey Sanders: Right. That's right. Absolutely, and we're seeing this, whether it be network connection, whether it be local proximity, whether it be security reasons, having this combined solution is something that, I think, we really invented with Azure and we're starting to see some of the other cloud providers actually come out with similar solutions. Although, I would still argue ours is both the best and sort of hardened, but ...

Corey Quinn: Well, let's not kid ourselves, of anyone who plays in this space, I don't think you'd be able to find someone who understands what it's like to deal with on-premise customer workloads for the past 40 years at Microsoft.

Corey Sanders: That's right. Exactly. We've got a little bit of history in this space.

Corey Quinn: Oh yes. Everyone remembers those days with a smile and a wince and it's ...

Corey Sanders: Yeah, exactly. Maybe a whimper, but hopefully, that experience is turning into something very valuable to customers today.

Corey Quinn: Well, it does. People think I'm being sarcastic when I say this, and I may have said it to you last year, but Microsoft has 40 years of experience in apologizing for software failures to customers because in the cloud, things break, computers fall apart. It's what they do. It's in their nature, and learning how to tell that story in a way to a customer that is first, sympathetic and also aware of the fact that they are in pain and not blaming them for it-

Corey Sanders: That's right.

Corey Quinn: ... is absolutely critical.

Corey Sanders: Well, and then always making the service better. This is the thing that I feel really passionate about. The opportunity to learn from both what customers are doing, how they're using our services and even the problems that we have and how to consistently make it better and better and better. That is one of the exciting parts of the cloud. In the history of on-premise software, it was a three-year cycle, three years later things got better and now it's three days later. The opportunity to sort of have that turnaround is really pretty exciting.

Corey Quinn: Yeah. The faster you can iterate forward in speed time to market, the more valuable it is for everyone.

Corey Sanders: Absolutely.

Corey Quinn: Increasingly, although not for everyone, of course, there's not as much business value as there once was in running your own data centers effectively. Let's not kid ourselves, if you can't run a data center more effectively than I can, with your resources versus my Twitter for pets company of four people, there's a serious problem for everyone.

Corey Sanders: That's right. Well, and let's be clear, it's not me personally, I'm not actually in there running it. I don't think you'd want that. You may actually be able to do that better than me, Corey.

Corey Quinn: Well, Microsoft solutions does feel a bit like a catch-all solution. We don't know.

Corey Sanders: That's right. They call me in when the plug gets pulled out, someone tripped on it and I plug it back in. But-

Corey Quinn: Then you're the hero.

Corey Sanders: I am frequently the hero. That's right. As any good Corey would be. Let's be honest.

Corey Quinn: Absolutely. It's all in the name.

Corey Sanders: Yeah, that's right.

Corey Quinn: It's our Corey competency, one way or another.

Corey Sanders: It is, exactly.

Corey Quinn: Microsoft, in general, and Azure in particular, have an awful lot of services. During the keynotes today, first Satya's, and then I got to see Scott's as well, it went from, "Oh yeah, that's interesting. I've heard of that. I've played with that, some. Oh, that one seems better," and then it sort of drifted into the realm of, "I'm not entirely sure if you were having a joke at the audience's expense or not," where there's so many service offerings, it felt like I had gone across the street to the Cheesecake Factory instead of flipping through their menu with all of the different options you can go through. It's almost analysis paralysis.

Corey Sanders: Like a New Jersey diner. Yeah.

Corey Quinn: Exactly.

Corey Sanders: Yeah, yeah, yeah.

Corey Quinn: It's overwhelming. And I've been using at least a few of these things for almost 30 years myself. From your perspective, again, bounding into Azure, what are the major tracks of Microsoft offerings?

Corey Sanders: Yeah, yeah, yeah. When we think about, and we talk with customers about it, we do split it up into two big categories. One being migrate, one being innovate. When you think about what, again, coming back to, what is the customer trying to accomplish? Are they taking deployments and they're trying to reduce some costs? Or trying to reduce some of the energy of maintaining it? Then migrate's your path and you're likely using something like infrastructure as a service and then a bunch of the surrounding services, security, identity, management to sort of make sure you can run that infrastructure in a healthy and clean way. Then there's innovate. A lot of the Build talk track is around innovate, for obvious reasons.

These developers were sort of building new things, but then you sort of have a little bit of the data side, and a little bit of the application side. Application side, we talked a little bit about two distinct services, app service and AKS, or Kubernetes service, and really focused on those being sort of the cornerstone of the application side of the house when it comes to innovate. Then, of course, data. Quite a few services on data. One of the challenges with data in general is just how many different types of data opportunities there are in the world, whether it be NoSQL, whether it be SQL-based solutions and then NoSQL there's like a dozen of different one to ... You find anyone on the street, and they'll tell you, "No, Mongo is the best." Then the next person would be like, "No, Cassandra is the best."

Corey Quinn: Then you have people saying snarky things like, "No, Mongo's great for your production data. Not my production data, that stuff's important, for yours it's awesome."

Corey Sanders: That's right.

Corey Quinn: It feels like the number one thing people love in that space is arguing about whether other people are wrong.

Corey Sanders: That's right.

Corey Quinn: They'd be, "One thing that's better than all else is being right when other people aren't."

Corey Sanders: That's right. That's why we do our jobs, so we can relish in that experience. This is where, some of the services that I think are really exciting, something like Cosmos DB where it ends up being multi-model. Cosmos DB comes in and says, "Hey, look, we're a NoSQL solution." You choose the model you want. You want Mongo, you want Cassandra, you want Gremlin, you can use it on top of this Mongo solution and it all works, and globally distributed, et cetera. It's really pretty, pretty powerful.

Corey Quinn: Do you think there's an architectural lock-in concern there?

Corey Sanders: Well, this is what's so beautiful about using those open-source models on top of Cosmos DB. You can come in and you can code to Cassandra, which is not locked. I mean, it's not locked in any way. You can go and run it in any cloud. You can run it on premises. In fact, we have customers who are running on-prem and Azure using cosmos DB as part of it, but you don't have to worry about the management. It becomes sort of, in my mind, the best of both worlds. You're not locked in, you've got this open-source model that you're using, but you don't have to worry about management when it's run in Azure. In some ways we're winning you over, hopefully, with the ease of use versus this sense of, "Once you deploy here you don't have any choice."

Corey Quinn: Right. One thing you mentioned a minute ago, there's a lot to unpack in what you just said.

Corey Sanders: I say a lot.

Corey Quinn: We'll take it piece by piece. You mentioned the build is aimed at being a developer conference.

Corey Sanders: Yeah, yeah.

Corey Quinn: Which is likely to raise an eyebrow or two from people in, basically, all of the tech cities that live on the coast that we all live in and we all know and love, in that, well, look at the customer stories you told. These were retailers, these were auto manufacturers, et cetera, et cetera, et cetera. I think that something that gets lost in a lot of the conversation is that IT or engineering, writing software, is not writing Twitter for pets in the middle of San Francisco where we've taken a job we can do from literally anywhere and build a land crunch in eight square miles on an earthquake zone.

Corey Quinn: Instead, it's now about things like a hospital in Duluth, it's an insurance company in Omaha. It's companies that are doing real world things that aren't just creating this new data manipulation or tying APIs together and calling that a service. These are companies that do things that have business models our grandparents might understand mainly make more than you charge. These are business model that our grandparents might understand where you make more money than you spend and that's called profit, which apparently is a dirty word.

Corey Sanders: That's right.

Corey Quinn: It's neat to see developers who are writing "enterprise" software are not being forgotten, if anything, they're celebrated...

Corey Sanders: That's right. Yeah. I mean, I think that's exactly right. Especially, when you look at the breadth of different customers we had up there. Obviously, had a lot from Starbucks, then Virgin Airlines. The breadth of these types of customers, and the problems that they're solving. One of the things that we talk about a lot inside the company is every company in the world is becoming a software company, because every single company is now thinking about, what is the software that we need to build to be able to deliver services to our customers?

Whether that be retail, whether that be manufacturing, whether that be financial services, they are all building and developing solutions. That's why a developer conference includes folks from all of those industries across the world. It's a very exciting time to be in technology, frankly, because you're just seeing this really blossom no matter what industry you're in. It's all about the tech.

Corey Quinn: Absolutely it is. Although, I do question the validity of some of those demos. For example, you had Starbucks up there doing a whole demo talking about what they were doing with Azure, and they didn't mispronounce a single service name. They're Starbucks, getting people's names wrong is their entire take. I have to wonder how many takes it took to get them there, "No, no, it's called Azure. No, that's not how you pronounce it. No, wrong company. Try again."

Corey Sanders: Now remember that, part of that problem may be the handwriting on the cups. Maybe when someone else wrote it, maybe they didn't have the same challenges that they have inside their stores. So that could be, "Maybe we figured something out here."

Corey Quinn: Cache invalidation, naming things and renaming things. Hard problems.

Corey Sanders: In fact, that's right. Coming up with new names for sizes is definitely something that they needed a machine learning model for.

Corey Quinn: You talked a bit about Azure serverless databases as well, on stage today, you collectively. Which was fascinating, in the idea of pay per second in return for performance, so we can scale down to nothing.

Corey Sanders: That's right.

Corey Quinn: Out of curiosity, if something is stopped, and you just start it up again, is that going to have a cold start issue? Is that going to just suddenly be there and ready to go through some sort of interesting caching layer? What's the story there?

Corey Sanders: Yeah, the thing about both of the serverless product than the hyperscale product, both announced today, very, very exciting, is they effectively separate the storage from the compute. And then it allows a lot of things to happen. One on the hyperscale side, it allows you to scale horizontally as needed. When you look at some things that you'd normally would be concerned about based on how much data you have, like taking a backup. Normally, when you're thinking about a database, you take it back up, you're suddenly like, "Oh gosh, how much data do I have? How long that's going to take? Is it going to slow things down," et cetera.

With the new ability to split and scale, suddenly that becomes a non issue. Similarly, with the serverless solution, it allows you to effectively separate out your data storage from what's going to run on top of it. Things like backup or things like computational queries and so on. It allows you to split them apart and run them as needed. There won't be a cold start problem per se because it is still compute nearby, but the key thing is the separation and then the scaling as needed. You can scale both tiers independently, which you think of a classic horizontal, excuse me, vertical database. You have that sort of scaling problem where one may scale more than the other.

Corey Quinn: It also allows you to scale without downtime.

Corey Sanders: Yes, that's right. Exactly. Exactly. One, on the hyperscale side allows much, much better performance, and on the serverless side can result in much better cost efficiency.

Corey Quinn: There was also an announcement today about aspects of your database offering being able to run at Edge and on ARM, which is fascinating. Can you tell me a little bit more about that story?

Corey Sanders: Well, this is the ability to take SQL and run anywhere, I think is a key aspect. One of the things that we launched, what, two, three years ago, was SQL running on top of Linux, which is a big step forward for the product. There's taking that product and be able to run in any form factor, being able to run really, really small, being able to run really, really large, being able to run on Linux, being able to run on Windows, that opportunity to take SQL, it makes it much more portable. It allows you to avoid, again, this lock-in point. You can take it anywhere you need it. You can take it and put it on an Edge device, you can take an and deploy it in the cloud. It'll work anywhere you go.

Corey Quinn: It seems like it's going to be unlocking a lot of interesting stories. Do you have any customer stories you can relate today, or is that still too early to answer?

Corey Sanders: We have a lot of really exciting Edge-based stories, maybe not as many today yet that we're ready to talk about, using sort of this new database offering with ARM. But quite a few examples where we have people doing super interesting things with Data Box Edge, being able to do computation on the Edge, being able to take sort of information running from drones and take cameras and being able to sort of do visual representation of those cameras. I think we demoed that last year as built-in Edge-based functionality on those IoT devices. There's a lot of opportunity here, and we're really just sort of scratching the surface at this point.

Corey Quinn: You also have live-announced today an early preview, a Bata Box Edge Heavy, I didn't catch the exact numbers other than 650 pounds, which is a strange way for me to measure data storage, but I'll take it. What is the story there?

Corey Sanders: Yeah, I mean, we had Data Box, which ends up being a way for you to take data, ship it to us. We have folks who are doing that, and it's portable, you can pick it up, a human can pick it up and it's ruggedized. You drop it, it's not going to be a problem and so on. But then we've now added this Data Box Heavy solution, which is one of the more interesting aims, I guess, we've chosen, which is not something that a human can pick up. It ends up being...

Corey Quinn: Well, not with that attitude.

Corey Sanders: Yeah. If you get in the gym more, maybe.

Corey Quinn: Exactly.

Corey Sanders: But it's just a significantly larger device for storage and being able to basically transmit a ton more data right into the cloud. It ends up being a great opportunity for other solutions where the network isn't as rich or as open or just the amount of data's so prolific that it just takes that type of device to bring it up there.

Corey Quinn: How much data does that 650 pounds actually let me transport?

Corey Sanders: It allows you transfer up to one petabyte of storage and it's secure, so it ends up being encrypted as part of that transfer. So a lot. That's a lot.

Corey Quinn: Right. Not only is it going to be encrypted so someone can't steal data off of it, there's no way any reasonable person's going to be able to even lift it in the first place.

Corey Sanders: That's right. The one petabyte actually weighs a lot apparently.

Corey Quinn: Yeah. Absolutely. It's-

Corey Sanders: Is that not a measure of weight? I don't know.

Corey Quinn: It feels almost like it's a mind bender. How much does a petabyte weigh?

Corey Sanders: That's right.

Corey Quinn: And I betting that it's decreasing line over time.

Corey Sanders: But is it in a vacuum or not? I don't know. It's like-

Corey Quinn: That would be mass not weight.

Corey Sanders: That's right. A petabyte of feathers versus a petabyte of stones. I think it's still a petabyte, either way.

Corey Quinn: I think you're right.

Corey Sanders: That's right.

Corey Quinn: Something else that was released-

Corey Sanders: We've done some really deep thinking here today.

Corey Quinn: We really have.

Corey Sanders: It's been lovely.

Corey Quinn: Truly. Something that was mentioned about a month or so ago as best I can tell is the premium pricing for Azure functions. Specifically, you pay a little bit more on a per function basis and in return you don't get cold starts. Things are pre-provisioned, which I'm of two minds on. First, it sounds great that you're able to pay a premium for a tier of offering that doesn't have a cold start problem, and suddenly it's right there ready to go whenever something hits it. On the other, it does irritate me a bit to hear people devolve ...

So the serverless function discussion down into always talking about cold starts or not, because there are so many interesting use cases for this sort of thing that go well beyond someone is clicking in a browser and watching a spinner go until the site finishes loading. For that use case, yes, absolutely. This is awesome. Based upon what you've seen with people adopting Azure functions in the wild, are you seeing that those are the primary use cases? Are you starting to see people use these as more backend processing, where if it takes an extra couple hundred milliseconds to spin up, it's irrelevant?

Corey Sanders: Right. I would say, we are seeing a lot of both. I think this is why actually having the two offers is so important for the customers who, to your point, don't really care about that immediate reaction time and really don't need it to be warmed up when someone clicks on it at all times, having that sort of ability to just react and do it casually, casual functions, as it were, which is not the product name, but should be, I think. The backend processing, there's a lot of things, even response to IoT-based actions, where there may not need to be an immediate response, some sort of signal or some sort of information that that has popped.

Given there's typically a large amount of time just to get the information to you, a couple, like you said, even a couple of seconds is not going to actually make a big difference, but in the cases where it is a human interaction, UX clicking on buttons, et cetera, where it actually will change, fundamentally experienced, by adding one or two seconds, then you can pay a little bit of a premium and basically have that warmed up.

From our side, it's really just comes down to, "Are we starting it cold, quite literally, from nothing and we're just going to go spin up that function, or is it ready and waiting to run and just perhaps taking a little bit less of that processing power?" That allows us to have that different approach. I mean, I think we are seeing a lot of both. Certainly, a lot of the cold start is being used today, and I'm excited about this offering, because I think we'll see a lot of that warm start now come through.

Corey Quinn: One last topic that I wanted to cover before calling it a show, is the idea of lock-in. People talk a lot about it, and often in some of the stupidest imaginable terms. One of the undercurrents that I saw, through both keynotes I saw today, has been a repeated effort for Microsoft to reassure people that lock-in is not a concern. Everything is open, whatever you put in, you can take out. To me at least, it always seems like a bit of a red herring concern, because even if you build everything in the most open possible way where you can take it anywhere you want in a day, great.

No one is going to move $50 million worth of resources overnight anywhere. Plus, dealing with staff retraining or turnover as a part of that. Plus, while that's in flux, what happens? You can't generally take a multi-week outage, for most use cases, to do a migration. Everyone talks about going significantly out of their way to avoid lock-in, but in practice we're already locked in, and in many cases, by our own data gravity, by the choices we make around technology. How does Microsoft view lock-in?

Corey Sanders: Yeah. It's very interesting, I think, when you really sort of dig into customer motivations and customer experiences, there's probably two different tiers of the way customers approach lock-in. One is at the infrastructure level and, to your point, there's a little bit of, no matter what. Customers are locked into their own premises database today in many cases. They've got the tooling, they've got the powershell scripts that work, they've got the processes that work. And so, even moving to the cloud is breaking from that lock-in.

Customers wouldn't probably call that lock-in, because infrastructure management is infrastructure management. It's going to change and you're going to have to learn it no matter where you go. And so, to your point, there's sort of a baseline set of challenges just to move anything, period. Now, obviously, we want to try, and make it as simple as possible. Solutions like Terraform can enable sort of a little bit of that mobility, but there is still a management aspect. There's a monitoring aspect, there are going to be some deltas, even between clouds, even using something like Terraform to sort of offer a layer above.

Corey Quinn: And you're still looking at that point only using baseline primitive services. The higher level platform offerings are never going to be one to one compatible.

Corey Sanders: That's right. This is where, I think, when I think about lock-in and some of the approaches that we've taken with our platforms to minimize this as much as possible, the open-source capabilities that we have on top of our past services dramatically improve your ability to move if you want to. I think this is really where when I talk with customers about lock in, it's less around, "I want to move every week back and forth and back and forth and back and forth." No, it's more, "Hey, I want the option to move if things go sideways, if it turns out negotiations don't go well or if I don't like you anymore." Hopefully, that's not true for us in any case, but the opportunity to say, "And I don't want my developers to have to completely rewrite the app in those situations.

So yes, it's going to take work. It's going to be a migration cost. It's going to perhaps be a period of downtime to move things, but I don't want to have to go completely rewrite things." When you look at some of the sort of classic examples of lock in that really gets customers frustrated, and I'm not going to mention those customers by name here, the biggest concern is, "I wrote an app, that developer has moved on and now I have nothing that I can do to fix it other than perhaps hiring a whole bunch of new developers for." When you think of like Cassandra support on Cosmos DB, Postgre support on SQLDB, Mongo, all those examples, they offer this portability that is an escape hatch. That's actually really important for customers.

The ability to say, "Look, if you guys really screw the pooch on this, we have the option." And that is why I love those open-source capability. Even AKS, Azure Kubernetes Service, I love those open-source options because something like AKS, fully downstream compatible. If you love our service, stay. If you want to use our serverless capabilities, stay. If you decide you actually want to move to somewhere else, it's fairly easy to be able to just say, "Great, I don't have to completely rewrite everything. It's going to be a very similar experience." There's variances to this lock-in point. I think our open-source focus for a lot of our platforms, and then portability focus, even SQL is very portable, gives customers this option if they need it.

Corey Quinn: Yeah, it definitely seems like there's a long history of learnings that have helped shape a lot of the decisions that Microsoft has made.

Corey Sanders: Indeed.

Corey Quinn: Just from talking to customers, seeing the pain and the suffering and the triumph and the tears of running infrastructures in the 90s and the early 2000s. It's strange in that, first, that's an incredibly valuable learning field. But, secondly, I've got to say, I don't recognize the old Microsoft in what I'm seeing coming out of you folks today. I think that's a compliment, but-

Corey Sanders: I will take it as a compliment, whether you meant it that way or not. Yeah, I know. Going back to the beginning, even as you talked about sort of our vision, it is a very new style of vision that really all we focus on is how we're enabling others. It's very exciting. It's a great time to be in tech and a great time to be with Microsoft.

Corey Quinn: It certainly sounds like it. If people care to hear more of your wise words of wisdom, where can they find you?

Corey Sanders: Oh gosh. Twitter. If you want to hit me up on Twitter at Corey Sanders WA, W-A. That was originally Windows Azure and then we rebranded and I didn't change my name. Then it became Corey Sanders, Washington state. Then I moved to New Jersey and now I don't know what it is. So Corey Sanders WA on Twitter and questions, comments, people hit me up. We have a show, actually, Tuesdays With Corey Show that you can find there as well.

Corey Quinn: Wonderful. We will do our best to come up with a backronym for that Twitter handle these days.

Corey Sanders: Yeah. Please, tell me what it should be and I'll change it.

Corey Quinn: You heard him, Twitter. Thank you once again. Corey Sanders, corporate VP of Microsoft solutions. I'm Corey Quinn and this is Screaming In The Cloud. 

Announcer: This has been this week's episode of screaming in the cloud. You can also find more Corey at screaminginthecloud.com, or wherever fine snark is sold.

Host: This has been a HumblePod Production. Stay humble.
Newsletter Footer

Get the Newsletter

Reach over 30,000 discerning engineers, managers, enthusiasts who actually care about the state of Amazon’s cloud ecosystems.

"*" indicates required fields

This field is for validation purposes and should be left unchanged.
Sponsor Icon Footer

Sponsor an Episode

Get your message in front of people who care enough to keep current about the cloud phenomenon and its business impacts.