Bare metal vSphere on Azure - CTO Advisor 069

Machine-generated from the episode audio and not hand-corrected, so names and technical terms may be imperfect. The audio is authoritative.

Hey, how's it going? com and we're returning to our regular format. We have Mark back. Mark, how's it going? I'm doing pretty well. How about you, Keith? How you been? You know, I cannot complain. I saw you under circumstances. I wish I would have seen you on better circumstances, but we are very happy to have you back. You mean when you creeped into my house and woke me up at three in the morning? I did make the bed up though.

You did. Completely true. That is the, you know, so there's that. And that other voice on the line is our good friend, Tim Carr. Tim, welcome back to the podcast, man. It's been, have we podcasted with you this year? Yeah, I don't know if it's been this year. We've definitely done it in the past, Keith, and I've definitely not been on the same line with you and Mark May at the same time. So God help us all.

This could get ugly really quick. Hands on the mute button, boys. You know, the great thing about this podcast is it's not done live. We have that going for us. So you're listening to episode 69 of the podcast, we're sponsored by Druva again, and thanks for Druva for sponsoring this episode of the CTO Advisor. Druva delivers the industry's first data management as a service, a single SaaS platform that unifies data protection and management for endpoints, infrastructure, and cloud applications. Unlike traditional systems, Druva aggregates this business critical data for scalable backup and disaster recovery, while also unlocking the value of search and advanced analytics for the governance of that data.

For administrators, Druva means simplification for IT leaders, scalability for CFO, saving a boatload of money, InfoSec, security by design, by the highest standards, and for VMware Nutanix environments, which we'll get into a little VMware Nutanix stuff, Druva means unified backup, disaster recovery, archival, all powered by AWS. Druva roughly translates to saving your butt from data loss, litigation, mishaps, and regulatory fines. Druva, the future of cloud data protection. All right, with that out the way, let's talk, you know, I'm glad nothing really happened this past week in IT, you know, this is not, you know, we're most definitely not a newsy type of program, you know, we usually talk about topics that you can go back and listen to the past few podcasts and say, oh, I don't know when this was recorded, because it's just a generic topic about how do I run and operate my enterprise IT infrastructure in general, what's the opinion on the overall trends in the media, but this week, man, the Twitter sphere, if you don't listen to Twitter, the Twitters were all the rage.

Mark, can you give the folks a background of why that's the case? Yeah, you know, I was actually gone for about a day when this happened, and I came back to all this crazy noise, and not the kind of noise that we bring actual, you know, kind of good noise, real noise, right, real noise about Azure offering a VMware on Azure service, which, you know, kind of flies a little bit in the face of VMware cloud on AWS. And I don't think VMware took it very well.

I was a little surprised. I don't get the impression that VMware did take it very well. Tim, you know, this is basically, you know, this is basically VMware cloud on AWS. This is this equivalent, right? Well, I mean, I think we're going to have to wait for the, you know, the market to play itself out. Looking at the announcement, it looks like, you know, there's a Microsoft has gone out there and said, okay, we have a partner, and I want to put the partner in like air quotes, right?

Like this partner is going to assist with deploying the VMware environment on what would be known as, you know, what we think is Azure hardware. That's all we can pull back from the announcement, right, from Microsoft's announcement. You know, beyond that, you know, there's just so much that goes into this, right? Like, clearly, you look at, you know, vSphere on AWS, and this is like an architected solution of partnership between VMware and AWS directly. This is something different, right? You know, Nutanix did this similarly with Cisco and HPE.

On paper, if I install KVM on x86 hardware, what's the big deal? The, you know, it's Linux on x86. That's kind of the way things should have to work. And so Nutanix made a big splash about being able to support their AHV hypervisor and their hyperconverged platform on these two partners, quote unquote partners, who are not really partners. They're just, you know, it's x86 hardware. I install my software on it, and I'll run it. Yeah, what's the big deal?

And then the same case with VMware and Microsoft. Microsoft, allegedly, has had HP ProLiant servers in their hyperscale cloud, so I'm installing vSphere on this. So, Mark, what's the big deal? Where's, I don't see the, well, I'm saying this, you know, tongue-in-cheek, I don't see the big deal. Where's the support issue? You know, I don't think there should be. I think the technical challenge is always the network, which I think is the same thing you would see with VMware Cloud on AWS, is encapsulating that network inside your existing cloud networking.

That's probably pretty hard. I haven't done it, but I think we can all assume that's pretty hard. But other than that, from a support perspective, I don't think it should matter. I don't think it should matter at all, but where it might matter is, of course, from a financial perspective, how this competes with another offering that they have looked like. Yeah, I'm trying to understand what they actually meant, like, from the existing, you know, from the Microsoft announcement and from also the VMware kind of announcement from this.

So, when VMware says, you know, they're partnering with AWS, that means, you know, technical resources from VMware, technical resources from AWS getting together to understand, you know, how they're going to solve the challenges, you know, to Mark's point, like that network encapsulation bit, how you're going to solve challenges like the, you know, storage and availability capabilities, right? From the Microsoft announcement side, it's all about, you know, when you say partner, VMware partner, you know, it can be so many different things that, you know, there's different levels of, you know, integration partners, there's different levels of, you know, sales partners.

I mean, like, there's so much gray area there. And I think to that end, that's why we have so much of this, you know, like this blow up that we see in the Twitter sphere right now, right? Like, I think there's a lot of questions to be answered around, you know, just the two sentences that went into that release, right? So, interesting. Yeah. So, I'm headed to AWS reInvent this coming week. Tim, will you be there? Oh, I'll be there.

Yes. Oh, well, you know, we both live in Chicago. We only see each other in Vegas, which is odd. It's sad, but true. It is sad, but true. So, I'm expecting, I'll get briefed by Microsoft, actually, when this podcast is published, I would have been briefed by not Microsoft, but VMware on their AWS announcements for the show. I'm expecting some, I have some embargo release information, and VMware is going to do some cool stuff. I'll just leave it at that.

Let's go into the network piece by itself. I alluded to this in the CTO Daily Dose on Friday, which is, basically, there's no layer two in most cloud. Like, when we, when I say layer two, there's no ethernet. That stuff really doesn't scale in a hyperscaler. So, they've gotten rid of that. However, vSphere is built, vSphere networking is built on layer two, like NSX. You can do stretched layer two to VMware's cloud partners, all this cool stuff that allows you to do legacy IT inside of a virtualized environment.

Now, I take that virtualized environment and meld that with AWS, who has no layer two at all, and Azure, who has no layer two. When I first heard the VMware on AWS announcement, Mark, my thoughts went just to yours. The networking alone has been difficult. So, Tim, without having, you know, a crystal ball and, like, real technical information, what are some of the top concerns or considerations that come to your mind when you see an announcement, bare metal VMware inside of a hyperscaler's cloud?

Yeah, I think the networking clearly is a big deal. You know, I briefly touched on something else, you know, a little bit earlier. I think, you know, just availability constructs, right? The networking in VMware kind of begets your capability to, you know, recover in certain scenarios, right? Be it, like, in a cluster or, you know, even as you kind of advance, maybe even as you kind of think of recovering from a site to site, right? Networking is crucial in those capabilities.

From a recovery perspective, having a network that's responsive and allows the platform to understand how it's going to recover is key. And I think that that, in my mind, is my single largest concern. And that concern isn't today, for me, really resolved in the AWS offering. I know that they're doing a lot of stuff around that. You know, one availability zone in that AWS offering today, they're working towards making that better, right? And I think that, you know, and keep in mind, in that AWS offering, that's with VMware working hand-in-hand with the AWS technical staff, right?

And so, you know, yeah, I have to say a lot of concerns around availability in my environment, and then when I have a challenge in my environment, recovery of those resources. So that's my largest concern. So when you're talking about availability, though, if you're looking at a classic IT shop, you know, they probably only have one availability zone anyway, or two or one active. They have individual availability zones. Sounds a little bit like you're trying to apply that cloud-native construct to legacy VMware.

You know, in a lot of ways I am. So here's, let me take a step back and kind of get to the heart of what I'm talking about. So let's just take one cluster within a VMware environment, right? When that cluster's running, you know, availability is maintained through those cluster nodes talking amongst themselves. You know, this is very high level, and there's a lot more minutiae involved in this. But there's a lot that happens from a networking perspective that allows those nodes to understand that, hey, everyone's healthy, we're all good.

In the event of like an isolation or in the event of like kind of a network segmentation, those nodes are kind of programmed to understand how they should respond and consequently how maybe a VM that was running on one bit of your cluster is then re-enabled in another. What I'm getting at is, you know, if you're just going to take this thing and slam it into hyperscale cloud, there's a lot of networking constructs that you have to have covered off on day one, to Keith's point, just to ensure that those basic HA capabilities are available in your environment.

So I think, you know, just step one, can we just do native vSphere HA? And, you know, I kind of wonder. So, you know, especially to Keith's point, you know, the very limited networking capabilities that you would have, or I shouldn't say limited, I should say different networking capabilities of a hyperscale cloud. Yeah, so there's also the other part of this announcement that I wanted to tease out was Microsoft was very careful in saying, in positioning this as a migration service versus, you know, the marketing around the VMware cloud on AWS is very different.

This is positioned kind of, hey, you need help getting your stuff to Azure, this is the way to do it. Like, I've heard industry people say you can use VMware cloud on AWS to transition to AWS services. I haven't heard any VMware people say that. What's your, either one of you take, you know, take a stab at this, what's your take on that? Is this just, you know, is this more smoke and mirrors and kind of limiting the expectations of a customer saying, oh, I can just get rid of all of my vSphere and run it in Azure, you know, indefinitely versus, because that's what VMware cloud on AWS is, and this Azure service seems, I think they named it Azure Migration Service.

Is that different? Yeah, I think for me, Keith, the thing that I was actually most interested in in the announcement, I hate to be this way about it, you know, I look at the VMware aspect of it, I'm kind of like, oh, Microsoft's kind of doing a B2 kind of thing here on the VMware side. I was actually most excited about the migration service that they were talking about. Like, when you dig into that, when you kind of peel back the onion there, I think there's a lot to like about that, right?

When I'm working with different organizations, you know, trying to help them kind of re-platform in the cloud, being able to map applications, being able to understand dependencies, and then being able to retrofit those sorts of applications into kind of public cloud constructs is by far the most complicated part of, you know, the technical side of bringing applications over. So let's tease that out a little bit. What do you mean by that? From a practical, give us a practical example. Yeah, so, you know, if you're going, you know, I have, you know, for example, we have, you know, maybe a, you know, customer that wants to be able to say, hey, I'd like to take this suite of applications or, you know, these, you know, this data center and move it out to the cloud.

Well, you know, when you say that data center, you know, average IT shop, you know, you know, we're talking hundreds, if not thousands of applications that are kind of running in there, right? And probably all kinds of different things. Yeah, by the time I do, so a practical problem, by the time I do an inventory, because none of our CMDBs are up to speed. We love it to be right now. We love it to be, but once I complete a cloud ready assessment for the thousand applications I have in a single data center, it's already out of.

Yeah, just still up 10 more by the time. It's already out of date. So the practical, just a practical challenge of saying what, what applications are cloud ready, quote unquote, do I have in my data center? That's for the average IT shop, a nightmare. A hundred percent. And that's what I like about Microsoft's messaging around this is, you know, I've not been shy about the fact that VMware cloud on AWS, I don't think makes sense for a lot of the workloads, unless you're just making that checkbox or you really need adjacency to AWS, but a way to, you know, take your applications, make sure they are cloud ready and move them to the cloud.

That marketing message resonates a lot better with me. So Tim, let's go back to that process. So as a migration service, if I'm using this as part of my migration strategy on surface, I can take, let's dumb it down real low. I don't know. Some people are going to be offended by what I say, but you know, just, it's conceptual. I've v-motioned my applications from my existing data center as is into this staging area. That's this bare metal VMware vSphere in Azure.

That, that's how I read the announcement. Is that similar to how you read it? Yeah, that's similar to how I read it. And then they're going to give us this, this set of tooling around that to kind of help us identify, you know, the, the problems that might come about as a result of that. That's kind of that migration service component that I read about it too. And as I, and again, we're going off of a lot of information that we don't have, and I'm talking from a customer perspective, what I would like to see in this and Mark, I'd love for you to jump in to this as well.

Cause you mentioned VMware cloud on AWS makes sense when you need, when you have adjacencies from your vSphere environment set of applications to your Amazon stuff. So I want to run a legacy Oracle app, not get deemed for the Oracle licensing in AWS. Plus I like bare metal Oracle. That makes sense. At some point I may decide to move that to my SQL. The advantage of having it in the same data center is that one, the, I have adjacency networking perspective, so super low latency network, but I also have the ability.

If, if it doesn't work, the fail back is really easy comparatively in the same data center. If I try and migrate my Oracle app to my SQL in the, uh, on, on, in AWS doesn't work. You know what I hit the undo button and I'm right back in my bare metal metal Oracle or bare metal vSphere running Oracle in AWS is data center. I'd see that appeals to me. Uh, when I read the migration services in having tooling specifically to move it to Azure, tell me what type of tools would you like to see?

Yeah, I think anything that gets us closer. So this is kind of a phased approach, right? I think, you know, there's a lot of, there's tools out there in the market today that'll just help you go directly from your data center to the cloud. Some of the limitations of those tools today are, you know, how do I get back to my data center to your point, Keith? I think as we get closer and closer to, we can just pick this thing up and run it in a cloud.

I think the tooling is going to have to get better at, you know, first of all, rationalizing dependencies. And that's what I like about this announcement. It's got to be able to do that. Number two, it has to be able to, you know, potentially help us understand what a light refactor looks like. So is our application going to remain, you know, like over provision over, you know, kind of like we would normally do on our typical on-prem VMware environment, you know, those kinds of costs are for, for any of those instances in like a, either a, in an Azure or an AWS environment, it's going to be pretty high.

So we need the right size, right? And so we need that sort of capability. I think finally, the thing that we're just starting to see, and I, and I think it's kind of, you know, as we go into reinvent, I'm actually excited to see about, excited to see this. I'd actually, I'm actually very excited to see anything that's going to talk to me more about the data layer, right? How intelligent can you be about analyzing my data, my actual data layer and telling me, you know, can I pick this portion up, refactor it into this database service in my cloud?

And what does that look like? Right? So I'm, I'm excited for tooling that's going to help me go that way. And then give me a bundled cost number around it. That's Nirvana, right? But we're a long way off. Yeah, Mark, that, that sounds magical. Yeah, that does sound magical. And I like magic. You know, one of the things I struggle with in this whole thing, you know, I went to DockerCon about a month ago and a lot of the talk was around containerizing your application, or actually that's not what they said.

I call it containerizing your application. They, they, they call it modernizing your application, basically taking a monolithic application, monolithic binaries, doing what we would have called a P to V taking, you know, a set of binaries, putting it into a container and running that container as a, as a service within a cloud. And now you have a modern application. That, that, that again is magical. This is hard work, Tim. I love your description of taking the data layer, doing some analysis to it, and then abstracting that data layer.

That's a, that's a serious, serious challenge of how do I take data from one application? Let's say SAP, very structured. There's very pronounced ways to interact with that data set. How do I take that data, which is what has value, put it in the cloud service and allow some type of API type of gateway to connect to that data and leverage that data and however I want it to be leveraged without breaking SAP. That's a real problem. That's a huge problem.

And I think until we get to a point where the migration services fix those problems, these are very, so let's put this in, you know, kind of perspective. These are very evolutionary, not revolutionary steps. This is just step one of let's put VMs close to the cloud and start to see what new capabilities or what customers are really asking for immediately. Yeah. And I think, I think that's, that's kind of part of it, right? We talked briefly around, you know, containers and what we're doing there.

And I think when you look at application landscapes and, you know, SAP is a very specific use case and, you know, you're right, Keith, you have very specific ways that you have to organize data, how data has to be managed within that environment. You know, that's, that's a really difficult conversation. You know, I would argue that that exact same problem exists in any of our, you know, kind of, any of our existing applications that kind of rely on these traditional, traditional database models, right?

Like we're not there in terms of being able with our existing enterprise applications that are relying on the SQL server or something like that. We're generally speaking, not there in terms of being able to, you know, just completely just say, okay, we're going to refactor this into something completely different. There are constructs out there that give us the capability to take all the other components, right? Like the app, the app components, you know, the web components, if it's a web application, even, you know, like the infrastructure services, like load balancers and, you know, security policies, there are components in the public cloud that allow us to do that, right?

But when it comes down to it, it all gets back to both, you know, the data structure and then ultimately data gravity within our environments, right? And so until we can begin to really start to answer that question, that's what I'm really excited to hear about, you know, at reInvent, I'd like to, I want to get closer to that side of the shop. Until we really hear about that, I think that, you know, it's, we're always going to be stuck in this middle land of where we're like, well, now we got to get this big, you know, honking monolith bit of legacy close to my cloud so I can take care of the other components with cloud services.

And we're kind of in that phase of the, of the shift today, I think. I think that's actually a great point in the fact that if we can't take a SQL database and make it MySQL or an Oracle database and make it SQL inside our four walls very easily, or I can't take the mainframe and make it x86 easily, I think data is always going to be a challenge. I don't see a way, at least right now, where anybody helps address that easily without a massive refactoring.

So cloud isn't magic, huh? No magic. No, sorry. All right. With that, on that, on that said note, Tim, you're obviously in AWS, at AWS reInvent next week. Why are you there? Who are you? Who do you work for? We didn't do that at the beginning. Why, why do we have you on? Yeah, so I'm heading, I'm heading out to reInvent. I am a cloud automation orchestration specialist with a little, little consulting company here in Chicago called AHEAD.

That sounds like magic. Yeah, it's still, it's, it's 100% magic all the time. I got to say, I love my job. I'm constantly, you know, talking, talking about demoing out, putting together different proof of concepts, working on these exact problems that we're talking about today in real life with customers. And it's, it's great. It's a lot of fun. I'm heading out to reInvent to kind of work from an AHEAD perspective. You know, clearly I got, I'm on booth duty, clearly I got, I'm on booth duty, number one.

And then number two, I'm really focused on the container space out there. So I'll be, I'll be over at the ARIA a lot. So if you're, if you're around there, that's where the majority of like the container sessions are. The majority of the kind of IOT sessions are. I'll see you out there. All right. And how can people stalk you online? Yeah, I'm at Timmy Carr on Twitter. And you know, yeah, that's pretty much the best way to get me.

I'm pretty responsive there. And Mark, if this is a, if people have been following Timmy to get to us, how can people find you? You can find me on Twitter as at Sensi storage. All right. And as always, you can find me on Twitter at the CTO advisor or at CTO advisor. You can follow the podcast at the CTO advisor on Twitter and don't be like Mark. I had to sneak into his house at three o'clock in the morning to set up his wife's pod catcher to record the CTO advisor.

You don't want that. Set it up for your mom and your wife already so that I'm not sneaking into your house to mess around with their pod catcher. Until then we'll talk to you next episode of the CTO advisor.