Transcript:
Episode Hosts: Pete Wright
Panelists: Kevin Jackson
Pete Wright:
Hello everyone, and welcome to Trilio Insights on TruStory FM. I’m Pete Wright. In an age where 94% of enterprises use cloud services, the way we manage our digital infrastructure has never been more critical. OpenStack with its vast capabilities for handling diverse workloads and Kubernetes the Go-to for container orchestration, both play pivotal roles in modern cloud architecture. Kevin Jackson, our director of product Management here at Trilio, joins me today to peel back the layers of these two giants in the cloud orchestration arena. Kevin, welcome. Hi.
Kevin Jackson:
Hi, Pete. How’s it going?
Pete Wright:
It’s going very well. I’m so glad to have this conversation and welcome you to episode one. I wonder if we could start with setting the table a little bit? My lens on these two blocks of cloud infrastructure is narrow. Can you give me a sense of just how important OpenStack and Kubernetes are to the modern cloud?
Kevin Jackson:
Absolutely, yeah. So I always like to start off in the world of history. So historically, OpenStack came first. So OpenStack is an open source cloud infrastructure platform, which allows you to basically have control over your data center resources. The focus is around ease of management of all those resources, of compute, networking, and storage. And so it is a great starting point for people to move off from the world of traditional virtualization. So people are managing virtual machines and that’s where the applications are run on. But OpenStack is an infrastructure as a service platform. It’s very much more geared towards the infrastructure management side of things. Enter stage left Kubernetes. Kubernetes takes the idea of being able to manage those resources, but you’re getting down to a much more granular level. You get into a stage where it’s very relevant for the application owners themselves.
So why is this all important? So from one side of the house where you’ve got IT engineers, they wanted a good way of managing their infrastructure. And then on the right-hand side, you’ve got the developing, very important for developers to be able to have consistency around the deployment and management of their applications. And so when you think about the importance, you’ve got one side of the house which is saying, “Hey, look, my infrastructure is important. I manage these data centers and I need to have an efficient way of managing those resources for my users.” And then the other side of the house, you’ve got the application owners and developers going, “Well, I want to manage my resources in an efficient and very scalable and an efficient way.”
Pete Wright:
My understanding is the modern way we make decisions like this is a cage fight. So we just put a developer and an engineer in a cage and see what happens. But what goes into making these decisions? If you’re put in a position of having to say, “Who am I making happy here?” What goes into deciding OpenStack versus Kubernetes?
Kevin Jackson:
Well, I mean, I guess that’s the interesting thing. I mean like anything in IT, I think one of the biggest elephants in the room is the fact that you can solve an IT problem in a thousand ways. There’s always a way of… You can manage applications just using the world of OpenStack. I can guarantee that the IT organization will be very happy because that’s how they can, as I say, easily, consistently manage the physical stuff. When you walk into a data center, they can see and touch things and they want to be able to manage them in an effective way. A developer could use that. OpenStack has some great tools. It’s got a fantastic ecosystem. And to be fair, if we’d had this conversation 10 years ago, people would be very happy to have a cloud platform like OpenStack, which will scale their applications appropriately.
It just so happens that somebody demanded a little bit more, somebody demanded more flexibility. Somebody demanded a little bit… Or looked at the limitations of a solely infrastructure service platform like OpenStack and said, “Actually, look, as a developer, I want to be able to control load balance resources. As an application owner I want to make sure that when I’m losing, I don’t know, a server or a data center, that I want to be able to make sure that my application continues to run.” So in terms of the cage fight, in terms of them versus us, I mean, this is where you get to that infamous world of DevOps. Whilst it’s not, this was never meant to be an answer to pure DevOps, and a completely overused phrase and term, it’s very relevant for this conversation. Because I do think Ops infrastructures a service, OpenStack and Dev, Kubernetes. And I think that’s certainly where this story should begin.
Pete Wright:
I was having a conversation with a developer friend, he’s an architect, the other day. And he said one of the things that he’s constantly trying to navigate is the budget of the unknown. And that often, even though we’re making a decision to go with an older tool in the stack, we’re doing it because the newer tool offers too many unknowns and it exceeds that budget. That we just need to go with the thing we know, the thing we have confidence in. Can you talk a little bit about making this choice from the perspective of the budget of the unknown? Is there something to that argument when making a choice between these two products?
Kevin Jackson:
So again, going back to that cage fight and the fact that this is versus.
Pete Wright:
Sure, I love being in the cage.
Kevin Jackson:
Ironically, we’re talking OpenStack versus Kubernetes. But ultimately without spoiling the narrative, there’s a complimentary force between them. There’s a natural force between the two. But we’re going with the narrative of this. So yeah, the infrastructure side of things, that’s their comfort zone. It’s all about being able to physically understand the fact that this is a compute node, there’s a virtual machine and it runs some stuff. To me, that’s a very comforting place for certainly a large swathe of the audience. But then if I was talking to a developer or an application owner, the whole premise of Kubernetes is the fact that they don’t want to know about the infrastructure. They don’t want to know about the IP stack. They don’t want to know about how the cables and the color of the cables and what storage is being used.
They only care that their WordPress application or their message queue, the Redis or whatever, is actually functioning correctly. From a conceptual side of things there are people who go, “Do you know what? I’m very comfortable with the fact that when I deploy my code, it just runs. That’s my level of comfort.” And then you’ve got the other side of the house where you’ve got database engineers going, “Well, I actually need to physically know which disc my tables land on. Because I need to make sure that those discs are spinning correctly or they’re available, because if not this whole organization will fail.” So yeah, I think there is that level of like, “Well, I know this world and I know this world.” And that’s why we have these two separate platforms.
Pete Wright:
It seems like your effort to break the dramatic narrative of this conversation has worked. Because you start bringing up this whole idea of management granularity, the idea that the code just works, but we can manage it at whatever level of detail we need to manage it. It seems like now we’re talking about not choosing one over the other, but how do we orchestrate them together?
Kevin Jackson:
Yeah, absolutely. So yes, I see it as an evolution. So OpenStack came along, as I say it was brilliant. It was meant to be the on-prem version of the world of cloud and Amazon and all the forces that came with that. So it was great to begin with, answered a lot of questions. But again, as I say, we’ve got this thing called Kubernetes now, which basically you start to abstract a lot of the concepts, which from an infrastructure person listening or if they’ve never seen Kubernetes, they’ll understand the concepts. There are load balances, there are this idea of code running on some say, worker notes. But it’s not that far removed to realize in the world of IT, everything we talk about here, there’s a physical server. Even though the world of serverless, which is a completely different conversation, there are still servers. So it’s just some servers. And all it is the fact that this code just has to run somewhere.
But somebody’s just taken this idea of instead of me saying that I am going to write this application and very specifically, it’s going to run on this group of virtual machines, this application owner is going to say, “Well, I’m just going to deploy to this particular cluster, and I’m going to let this cluster handle the placement of those particular resources. I’m going to describe in pure software, what happens when one of those pieces of software fails. I should have four or five instances of this code running on this cluster. This cluster is made up of many physical machines. I don’t care where it runs. I just need to make sure that there are five instances of this application running. I just need to describe that when somebody comes in and the request, it’s a low balancer, that it then ends up on one of these application nodes that are running. I don’t care what the name of the node is, I don’t care what type of technology the low balancer is. I just need to make sure that this is working as expected.”
This is where the element of control and the complimentary factors come into play with OpenStack and Kubernetes. Because OpenStack itself has this abstraction layer. In other words, I can wheel in some Dell servers and run OpenStack on there. I can run in some HP servers and run OpenStack on there. And people consuming that particular platform, they have no idea what the badge is on the front of those servers. So there’s a level of abstraction. You can define load balances. But it’s just a little bit more explicit. You’re actually telling OpenStack, “I want to create a load balancer, I want to create a virtual machine.” But in the world of Kubernetes, it’s just a case of, “I don’t know what you’re creating, but just run my application.”
I mean, to be fair, there’s a lot of people who know what’s being created. They’re creating containers or in the world of virtual machines in Kubernetes. But it’s a little bit more abstract, it’s a little bit more of just let the technology handle how things come to be. As long as I’ve described the relationships and the path of how things go from one place to the other, again, and the key thing here is impure software and then everybody’s happy. But this goes back to the complimentary side of things. This is where we have to think about the as a service layer in. So OpenStack is the infrastructure as a service layer. So, “Hey, I can manage my physical machines now in a programmatical API driven way.” Kubernetes is essentially an application that runs on top of OpenStack. Kubernetes has a little bit of dual functionality at this stage, because technically you can have Kubernetes managing infrastructure elements as well.
So this is where that blurred line and the bringing back this little cage fight of the versus comes into play. This is a case of like, “Well, I’ve got OpenStack and I’ve got Kubernetes.” Well, that’s really good because you’ve got the ultimate flexibility of managing physical hardware, physical data center stuff, as well as all the complimentary and all the abstract layers and the consistency that Kubernetes brings. But you can do away with OpenStack as well and just have Kubernetes run on bare metal. And from an application point of view, they seem no different. But the advantage of running one on top of the other is the fact that you keep the IT engineers happy, you can scale things seamlessly. And it’s that scaling and manageability point of view, which is why Kubernetes and OpenStack become very, very nice and very complimentary to work to each other.
Pete Wright:
Kevin, how does one go about navigating this blurred line? Everything you’re talking about says that Kubernetes and OpenStack can work together, and there’s a blurry line between the two. What do you do to bridge that gap?
Kevin Jackson:
Yeah, I think there’s almost like an overwhelming urgency to say, “Hey, look, I run virtual machines, but I need to go to this cloud native world of running printing containers.” And I think when we talk about the reason why a lot of people are in this middle ground of running infrastructure as a service as well as Kubernetes, is the fact that there are still applications that run really well in the world of OpenStack as virtual machines. And then obviously there’s this cloud native world, which begs itself to have a re-architecture, a redevelopment of their applications to actually work correctly with Kubernetes. But there’s a middle ground, in the Kubernetes world there’s add-on project called Qvert, so developed by the Red Hatters. So again, we see this in the OpenShift virtualization space of course, as well. But this allows you to run virtual machines on OpenShift and on Kubernetes as well.
And so what you get… And then this goes back to that point about people who can run Kubernetes on pure bare metal and just see that as the infrastructure platform negating things like OpenStack under the hood to do the scaling. But in terms of running virtual machines, it just allows that on-ramp into the world of Kubernetes, it allows them to use all the constructs and all the reasons why you would go to Kubernetes. The abstract layers, the fact you can describe everything in pure software. But you don’t have to refactor your application. It’s a virtual machine. You can onboard your applications.
And so you’ve got this way of managing your traditional applications and your containerized applications through Kubernetes, and it’s no different now to managing your virtual machine. So there’s a nice on-ramp, there’s a nice in-between. But I wouldn’t even say that Qvert or Open [inaudible 00:14:45] vert is a versus of OpenStack. It’s certainly complimentary. It’s just a different way of getting to the world of Kubernetes. But I still recommend that you would still have an infrastructure as a service under the hood, just so you’ve got that management layer for the physical stock. There are still physical machines you still have to manage.
Pete Wright:
It seems to me like what you’re describing is the circumstance of a mature infrastructure organization. Is that a fair assessment?
Kevin Jackson:
Yeah, I think so. Yeah, yeah.
Pete Wright:
So give me a state from your perspective where the industry is now. Are they moving in this direction toward maturity? What does the bell curve look like?
Kevin Jackson:
Well, yeah, it’s weird. I’ve been in this industry for quite a long time. I could say one thing about like, “Hey, this is what the industry runs.” And you suddenly realize you’re in a bubble.
Pete Wright:
Yeah, there’s your bubble.
Kevin Jackson:
Everything’s in a bubble. So in one particular bubble, the whole world runs Kubernetes. In another bubble, the whole world is running just a public cloud infrastructure as a service. But I think the reality is, I think there’s a big middle ground of people running traditional virtualization. There are people who are running pure OpenStack. There are people running pure OpenStack and dipping their toes into the world of Kubernetes. And then you’ve got some big organizations, and this is where the bubble comes into play, you’ve got some really big organizations that put 100% effort into the world of Kubernetes. Which makes it look like the whole of the IT world is running Kubernetes. But it’s just because it’s those big names. It’s the things which people standing up on stage at CubeCon, you think this is the be all and end all.
But I think from my point of view, what I’ve been witnessing is there is a large mature Kubernetes crowd. But I genuinely think that the vast majority of people are still trying to work out the best use of Kubernetes and the best use of something like OpenStack under the hood. And this is a very interesting world, certainly for Trilio, because we get to see both sides of the equation. But that’s where I think most people are. If we could get people to stand up and put their hands up and say, “Yeah, Kev, you’re absolutely right. This is where we are.” I genuinely think that it’s a mix. The biggest pool is a mix of people who are taking full advantage of OpenStack, which is really good. And there are people who are taking a lot of advantages from Kubernetes, but it’s not 100% all in one or the other.
Pete Wright:
You dropped Trilio and the things that you’re seeing in the industry. Since this is episode one, Kevin, I feel like I need to offer you the platform to talk a little bit about the Trilio participation in this space. Can you tell us a little bit about what we’re doing here?
Kevin Jackson:
Yeah, absolutely. Yeah, so Trilio provides back recovery and disaster recovery software solutions for OpenStack and for Kubernetes.
Pete Wright:
I’m shocked. I’m so surprised. I’m shocked. Shocked that there’s Kubernetes happening in this environment.
Kevin Jackson:
Yeah, absolutely. Yeah, who would’ve thought, right? And so yeah, this is the software which basically slots into the world of OpenStack and slots into the world of Kubernetes, and it’s the reason why we get to see both. And so from a personal point of view, it’s great because we do get to play with my favorite platform, which is OpenStack. I think anybody who knows me in the industry would actually know that I don’t shy away from that admission. But equally seeing what you can achieve in Kubernetes and just being able to work with companies who are taking advantage of Kubernetes or being able to scale our environment on both platforms. Yeah, I’m in a very unique and privileged position, to be honest.
Pete Wright:
Well, I sure appreciate this, Kevin. Thank you so much for being here, for educating us on how we navigate the world of our gentle cage fight. Thank you also for being our legendary first guest on the show. We sure appreciate you.
Kevin Jackson:
Thanks very much.
Pete Wright:
Thank you everybody for downloading and listening to this show. Thank you for your time and your attention. We encourage you to learn more, and I’ll tell you what the easiest way to do that, is just swipe up in the show notes for the episode. You will see links to everything that we have mentioned, all of the links to services and pages where you can learn more about everything that Kevin has been talking about for the last little while. On behalf of Kevin Jackson, I’m Pete Wright, and we’ll catch up with you next time right here on Trilio Insights.