41. Private Clouds are Coming Back – Tech Field Day Podcast

Private clouds are returning. Are they a new thing? Are they an old thing?

Are they an old thing rebranded as a new thing? Have we got the same capabilities on premises that we used to have in public cloud? Or is it all just somebody else's computer?

Even if it's in your own data center? Find out on the Tech Field Day podcast. Welcome to the Tech Field Day podcast, where we bring together a group of IT technical experts to discuss a single idea about key concepts in the industry.

This podcast features a variety of perspectives from members of the Tech Field Day community and is often recorded in association with one of our events. Tech Field Day is part of the Futureum Group, and this podcast is also published on our sister company site Techstrong tv. On this episode, we'll be discussing private clouds coming back in 2025.

But before the discussion, let's meet who's on today's panel. Start with Alison. Welcome to the show.

Hey, Alistair, Allison Klein with the Tech Arena And Justin Warren from my part of the world. Oh, hi Alistair. Yeah, it's Justin Warren here from, uh, pivot Nine Consulting And John Meyer in his amazing studio.

Hey, Alistair, John Meyer from Meyer Media, And of course, I'm Alistair Cook. I'm an event lead here at Tech Field Day, and one of the things that we see always in it is that cycle in, cycle out, everything that is old is new again, everything that is new is old again, and that seems to be something that's, that's happening as we see private cloud start to make a little bit of a resurgence in people's consciousness, sometimes driven by the idea that moving your successful production application out into the public cloud gets very expensive and maybe we want that deterministic cost back on premises because we're not ready for the operational and financial challenges of using a public cloud platform. Other times we just have this preference for running things on premises because maybe it's connected to some other part of our application.

Yet we want the agility and the mobility and, and all of the, the coolness of using cloud platforms. So as 2025, we're gonna see lots more private cloud adoption, a move away from maybe public cloud adoption. John, are you seeing customers, I know you had a big presence at uh, reinvent recently, which of course is all about public cloud.

Are you also seeing a lot of thoughts about resurgence in private cloud coming up for the next year? Alistair, the discussion of going back or not even going to the cloud has come up in topic here. Even at reinvent a lot of the customers that we were talking to, the cost is the number one factor that so many people are thinking about.

Like, Hey, we thought the public cloud was gonna be great, we're gonna move there, but we didn't realize that we had to refactor or the cost associated with it, like for like, is not what it should be or seem in the public cloud. You should build things for the public cloud using all the services versus trying to duplicate your data center in the cloud. Really, I think this is gonna be a hot topic here in 2025, is that they don't want the cost that's associated with it and understanding it.

So will we see even people coming back from the cloud or a slowdown of going to the cloud, I actually feel that we're gonna, might see a little bit of a slowdown from going to public cloud and might rethink about going there and staying with the private cloud. Yeah, I think we we're seeing some similar things, but I, I dunno that it's going to be a lot of repatriation of workloads. There's a little bit, uh, that always happens as people kind of look at what is this costing and do we actually want it to be in that location.

What we're seeing instead is, uh, that the, the growth of where do we place new things depends a lot more on where people have stuff already. I think something that's kind of flown under the radar and possibly why we we're now realizing that private cloud is a thing, is it's always been a thing. It's always been there.

Not everything went to the cloud. So really I think this is just people looking around and realizing that this surrounded by private cloud. We just probably weren't really calling it that before.

I actually think that John hit it on the head. I think that there's a massive, uh, focus on how can they drive some of the new applications that they want to deliver for their business, including, you know, expect expectations of growth of generative ai, uh, integration into enterprise IT in 2025. How are they gonna do that and how are they gonna afford it?

That Amazon bill is expensive. Amazon, um, itself is getting some pressure to help, um, companies bring, um, applications and, and uh, data back on-prem. Um, and I think that, you know, that's just a reflection of the sophistication of, uh, the it, um, uh, leadership to understand how infrastructure on-prem can help 'em with that bottom line.

Allison, I wanna jump in there. There's two things. Uh, Justin touched on it, why private cloud has always been there, right?

We haven't called it. We built those, uh, data centers or their servers in our network closet, right? That was over here.

That's our private cloud and that will always be there. And you, you touched on generative ai, so I've gotta touch just a little bit on it, is that we might be utilizing the cloud services for generative AI because the LLMs are so huge, we need the services, but then there's private cloud. We might run it there and pull it back because everything else might be too expensive or we still might wanna stay local.

Yeah, I think there, there is certainly a lot of interest in running the inferencing side of generative AI or any kind of ai. I'm, I'm actually not, not big on particularly generative ai. I think that's a fad and it'll probably go away once the money runs out.

Uh, but there is a lot of like machine learning and, and the old school style of, of ai that, that, again, that's always been there. Uh, a lot of this sort of fancy stuff's like you don't actually need a full LLM for it. You, you can do something with three linear regressions in a trench coat that runs perfectly well on your own site, uh, are particularly things where you're doing like vision systems in, uh, manufacturing that tends to run very close to the machines because well, that's where you need the system to be deployed and working, and that's always been true.

Again, I think it's a little bit about the terminology we use. The thing that has changed a lot, I think from, as you mentioned, John, when we built this before we were doing it probably old school and it wasn't really a cloud, it was just infrastructure. A lot of the techniques that came from public cloud have actually come back in-house.

And I think that, I think Alistair and, um, and Alison, you both touched on this, the sophistication of customers now has increased a lot and they're actually capable of running things in a cloud like fashion no matter where it's deployed. So we just call that private cloud. And I think that's, that's a really crucial element is, is what private cloud we can deploy now is very different to the private cloud that we thought we were deploying five or maybe even 10 years ago.

And as Justin says that that old private cloud was an infrastructure as a service cloud. If we were lucky, we could deploy virtual machines and get an operating system on them. But really when it came to the, what I see as the true value of public cloud, which is as a developer enablement platform, as a way of building applications rapidly, um, without having to reinvent everything every time we are seeing more maturity in, in those tools being available for deployment on premises or in public cloud.

It's a bit of a shift from the cloud idea that the way Amazon builds their website is the way everybody should build all of their infrastructure. com website. com website, but if it doesn't, that might not be the right collection of tools to build what you wanna build.

And so, uh, one of the things I've always looked at is what's on premises for enabling developers to assemble applications rapidly. And I think we're seeing a lot more maturity in, in things like, uh, Kubernetes and, and helm charts being used to deploy the building blocks rather than having a, an actual service that we consume that is that building block. It might be a database as a service or a queue as a service that's deployed as just a, a chart with a few configuration options that wasn't available to us when we first started building private clouds.

I think the technologies have changed quite a lot, but on premises we still don't have some of the fundamental characteristics in that this definition of a cloud. We don't have massive scalability. We don't necessarily have universal accessibility.

Are these things still really important for solving business problems with clouds? Is that mis definition actually that important for private cloud? Um, I think that there is a relativism in scale, Alistair, and I think that, uh, enterprises can pull off enough scale to make the concepts of cloud work for them.

Uh, the concepts of failover, the concepts of, you know, not not one node, uh, to rule them all. I think that they can make, um, cloud economics work for them. They can make cloud architectures work for them and they can make cloud native work for them.

So I I I don't think it's a problem. I think they can achieve it. Um, there's always going to be a continuum of how cloudlike structures exist in enterprise.

Um, I recently talked to Rebecca weekly over at Geico. If you haven't had a chance to talk to her and her team, please do. They repatriated everything, um, from public clouds into a private cloud, um, architecture.

They did it with OCP hardware. It helps when your, um, your VP of engineering happens to be the former chair of the OCP Foundation, but um, you know, they, they showed what you can do with infrastructure and, um, capabilities that are built, uh, for public cloud and actually apply them in an enterprise context so it can be done. And I also think that there is a right sizing answer in terms of the complexity of implementation that can be applied across enterprises.

So Allison, I like how you worded that, that it can be done and you could pretty much, I mean like what did we do before public cloud was a thing, we still got everything done. It might have been at a smaller scale. It's a scalability.

Can we scale out? Do we have the regional, the data center capacity, the global capacity? Do we even really need the global capacity for it?

And if you think about it, and here's a challenging question, I always like this and it might touch on the hybrid portion of it, but AWS in your data center. So think outpost, is that public cloud in your private cloud expanded into it? And now, um, really am I hybrid?

I mean what, so really you can use those services or those needed services and still scale out or you can still do it in your private cloud. Yeah, I've, I've long said that cloud is a state of mind. It's not a place or a location.

And I think that's, that's an interesting part because Allison, you touched on, uh, that, that it's enterprises who are capable of doing this. I think the way that cloud was adopted, we did see a kind of a two speed version of things. It wasn't quite bimodal as, uh, some people might remember if they're old enough, but it, uh, it did seem to split the market between enterprise at the top end and SMB at the bottom.

Those were the two kind of classes of customer that really adopted cloud hard. It's that big middle tier that had probably just a bit too much to do, like one application or two applications that really run the business and we'll just use SaaS for that or enterprise where we have the capability and the staff will build it all ourselves in the middle. You kind of don't have either of those things.

You've got too much stuff to do it easily, but not quite enough resources to be able to really do it well yourself. That I think still exists. And what, what are you seeing around that middle tier?

Like are they sophisticated enough or has it gotten easy enough for the s for the middle tier of, of, uh, organizations to actually do cloud themselves? I think it's a really good question and I think it really depends on is there something that is deployable for that mid-tier, the market off the shelf that doesn't require a massive IT staff to actually customize, um, the orchestration layer that, you know, how they're actually managing cloud services. I think that the technology has advanced so that office shelf solutions will work for them.

I I would be interested in your perspectives on that, uh, to see if, I think that that is the linchpin for the mid-tier of the market. I think that the technology has matured. I would love to, um, hear exactly where you guys think that is.

Now, Alice, are you asking if off the shelf products will work for them for their private cloud versus utilizing the services within public cloud? Yeah, I think, I think that one of the challenges earlier as for that mid-tier is that you needed a lot of sophistication in your it, um, team to actually implement private cloud. I think that technology has caught up and made that a much more achi achievable goal.

Ironically, the off the shelf stuff will probably be hosted in a public cloud and work for your private cloud. That's the, the, the fact of where everybody's going, where you don't need to install it in your data center. They do have off the shelf that you can install.

You can have your standalone version, but I think there are enough services out there or third party things that would support your private cloud that would suffice for your enterprise that already had a data center or still has a data center or private cloud, you know, stance. So it's that middle ground where organizations would benefit from having cloud-like capabilities on premises, but they don't want to run it themselves. So where the actual application runs on premises or some of their applications run on premises, but the customer isn't responsible for managing the infrastructure underneath it that's delivered as a cloud managed service to manage your on-premises.

And that's the model for the, uh, outposts from AWS, but I think that's probably a little too, uh, abstracted or not, sorry, not sufficiently abstracted for the, uh, the middle tier, uh, one of the differentiators I see is the, the small companies might use off the shelf software and so they're using, uh, off the shelf software as a service that are enterprise organizations for their line of business apps. They, they write them themselves or heavily customize them. And that middle space is where it's, it's some of the applications might be in-house developed or customized and a lot of them are, are off the shelf and that that's where that complexity sits in and, and gets to be difficult for those middle customers.

So being able to use a software as a service delivery for something that runs on premises may well be a, a really helpful way of running those. The other thing I wanted to hit on was, John, you talked about, um, scalability and these, these elements and, and then talked about what we used to do before cloud. Um, we moved much slower before cloud, we had waterfall software development.

We were waiting three months for a new version of an application. Uh, when we had a, a market shift, it took us six months to react to that. And that simply isn't acceptable when you're competing with somebody who's using public cloud service and using CICD pipelines has a full, um, microservices architecture where they can respond and deploy new features in hours to days rather than weeks to months.

And so life has very much changed since those days. Alistair, to touch on that, that's my other side of it. So while I think that private cloud will always be there and people, I don't think they're gonna be moving back in drones of it.

I think public cloud, that is the reason people go to it, is the speed, the deployability, the efficiency to you use new applications, use new services, they are there for the innovators. And that's what makes public cloud so attractive though, because I don't need to go talk to somebody to deploy it out. I can just go do it.

But that the, the biggest problem I see with public cloud is the cost. That would be the only thing that people want to come back. There's so much awesome stuff in a public cloud, but the cost is just too, so much, so big that they're like, whoa, whoa, whoa, hold on a second.

We, we gotta reevaluate this. But Yeah, that's, again, I, that's not really new. Like I, very early on in cloud, I, I, I coined the phrase of AWS is a a a cheap place to fail, but expensive to succeed.

And this concern about cost I, I think is quite revealing because that to me tells me that customers aren't seeing as much value as they used to. So if it's worthwhile, then you're quite happy to pay the money. If you need the scalability and the flexibility, then it's, it's great for that.

But we're certainly, where we are seeing people get a little bit sort of antsy about it is they're looking at fairly steady state workloads. Um, early on it was things like we just lift and shifted what we had before and we, we are doing what we did before, but we're now paying spot prices instead of infrastructure that we already own and is fully depreciated and is basically not really costing us very much money. That's where people I think are looking at it and going, huh, maybe we moved a little bit too quick into this or we, we didn't quite do it in the right way.

And that's where I, I just wanna push back on the idea of, of moving fast and breaking things as being the, the one true way that we should do it. The flexibility of cloud and the ability to change your mind I think is very useful, but that does mean that you can change your mind 600 times in a, in a, in a fortnight making terrible decisions a lot. So maybe there is some merit in being a little bit more slow and a bit more considered about, about it and thinking, well, do we really need to do it that way?

What could we do instead? And adopt things that people have already, like the people who are out there innovating have already gone and tested all of the things and they all set themselves on fire or broke their fingers or made hit all of the, the sharp edges. We can just sit back a bit and be a fast follower and uh, go figure out which ones are actually the good way to do things and do it once properly instead of six times the wrong way.

Justin, do you feel that, uh, the cost right is more of an education on utilizing the services that are available to you and analyzing what the cost is going to ultimately, you know, achieve over the long run of their application? Because you said where they were using a set steady state and using spot instances for some of their workloads. If they knew that going into it, do you think they would've moved to the cloud?

I think a lot of people did, uh, partly because of like, cloud is what we should be doing so we'll, we'll copy what everyone else is doing because well all, all of these people can't be wrong. And then they've sort of realized partly through learning how it actually works now they feel more comfortable about what they're doing. And that, I think to touch on a point that Allison made about, about the, the level of education that people have and their comfort with these tools, I think the tools have gotten a lot better, but I don't think the people and their understanding of it has caught up yet.

But it has gotten a lot, it is now a lot better than it was 10 years ago. The people understand how these techniques work a lot better than they used to, and so they're adopting some of these techniques more now. We need, we still need the tools to get better, but I think the big shift that's happening now is just with the teams of people who are running it, they're using terms like platform engineering to understand how do we actually construct this in a way that is easy for us to run and maintain with the proviso that we do want the flexibility of changing it when we need to.

Justin, do you think that they are looking at that from a standpoint of cost savings or from a standpoint of agility? I'm a consultant, so it depends. Uh, I'd say both and because it depends on what the application is and the sophistication of that team.

So speaking with, with customers only in the last couple of weeks, uh, some of what they're doing is, is they're doing a rethink of how have we been doing this and maybe we should reconsider. Uh, I get the impression that people are actually pretty tired. Uh, they've been scrambling and like trying to run as fast as they can for quite a while now.

And certainly post pandemic where people have kind of settled down, they had to do a lot in a short space of time. I think that workload of things that were shifted off because we had urgent things that we needed to take care of a few years ago, those are now really coming to the fore of, oh, we, we need to sort this out. So I'm seeing a period of reflection, uh, looking at things and reconsidering how it's done and then trying to, as you put it right, size things, put things in the right place and do it in the right way based on what we know now compared to 10 years ago.

Well, things I, I think is that, that we are getting towards this maturity stage of not saying it's cloud first. You have to have a reason for not being in cloud or it's on premises because we don't want to look at some of the complexities of working out whether we can be compliant. We're moving to a stage where we're organizations are saying, okay, cloud has a certain set of characteristics and it's beneficial public cloud for a set of use cases for places that we're interacting with our customers, where that massive scalability is important and that the elasticity, the ability to scale back in again and stop paying is important.

But there's a whole lot of our estate that is slow changing applications that align of business that may be a tightly coupled to on-premises physical, uh, operations or that have that, that flat line, that's that straight line kind of use case where the lower cost per gigabyte, per second of, of resources for on-premises when it's a sunk cost, uh, is, is more beneficial. I think we're seeing much more maturity of that combination of the two. Hopefully we're gonna see more of the ability to be agile whilst remaining on premises.

Returning to the idea of private cloud coming back and and stronger in 2025, hopefully we're seeing the tools that Allison is seeing, uh, that allow applications to be built on premises clouds more easily to move downwards into the smaller organizations, not require that you are a multinational, uh, financial organization in order to deploy on premises, but definitely seeing a move towards real cloud platforms on premises, some of that being driven by Broadcom's changes at VMware. I've certainly was, was looking at that in the rundown recently and and saw, um, maybe some moves towards other platforms. Maybe it's time for a resurgence of OpenStack or its derivatives in 2025 as on-premises private cloud.

I think that, um, we see interest in CNCF and, and what's coming out of cloud native stacks, um, for enterprise. I think that those are looking less scary than maybe they did before. And, um, you know, you can, you can expect that enterprises are looking at alternatives to VMware.

So VMware certainly doesn't have the, the lock in that it once did. And you know, when, when I've talked to enterprises that are repatriating, they're doing it with open source as part of their solution. So, you know, are we gonna start having OpenStax summits again?

I don't know. Uh, that would be fun. I like those events.

Um, but I, you know, I think that, uh, having open source alternatives is something that feels much more comfortable to enterprise than it did at the first time. We were talking about public cloud back a decade, decade ago, and that's probably some of the sophistication of the IT teams that we've talked about earlier. Are you allowed to put CNCF cloud native and private cloud in the same sentence?

They seem like two separate numbers for it because obviously the conversation here is the private cloud. Does this conversation only apply to those that haven't been born in the cloud and have to understand it because they don't know what private cloud is? Or will they have their own private cloud aside from the public clouds?

You're putting restraints, artificial restraints on us. John, I'll put in some challenges there, thinking about the word private cloud in the conversation that it's private and a public cloud. Uh, I don't know too many clouds.

I I missed the, the early definition days where we were all arguing if it really is a cloud and cloud. Watching that, that was fun. We should totally do that again.

Oh, the time for old being new again. Well, before we dive into all of the things that are being old, being new again, I think it's time for us to wrap up this conversation. We will of course carry on many more of these conversations in person at Tick Field Day events, particularly Cloud Field Day coming up in February in 2025.

Uh, but I'd like to thank you all for joining us today on the Tick Field Day podcast. And before we go, we can people connect with you and continue this conversation. Allison?

ai and, uh, you can find me on LinkedIn and all the socials. Uh, we're, uh, publishing multiple times a week on topics associated with cloud to edge computing. And John, where is the, uh, massive Meyer Media studio available?

All right, so Meyer Media is available at, at John Meyer everywhere on all social medias. YouTube, LinkedIn X, wherever you can think about it. Uh, powered by Meyer Media.

You can go to our website, John Meyer or Meyer Media co. All of the Meyer Media's everywhere. Yes.

Justin, how's your empire going? Oh, I don't, I don't feel the need to, uh, to have a whole empire. I'm, I'm quite happy just being a little boutique, the power behind the throne as we like to say.

com. Uh, I also encourage people to sign up for our weekly newsletter called The Crux. We're currently on, uh, on holiday at the moment, but we have a, a bumper holiday links edition, which is available for free for everyone to read.

Uh, check it out. There's a lot of fun links in there. And if that's the sort of thing that you are into, then please like and subscribe.

The most surreal part is when I was reading Justin's crs and, and, uh, particularly tickled by one of the articles and then described it to him when we were sitting down having a drink together and he pointed out, yeah, he, he'd brought the story to me. Definitely. I recommend that as a a newsletter.

You can find me, Alistair Cook Online is, uh, at deta nz. co, do nz as my, uh, personal blog and then I'm all over the place on LinkedIn and, uh, also up on the Tech Field Day and, uh, RUM group sites. So thank you for listening to this episode of the Tech Field Day podcast.

If you enjoyed the discussion, please subscribe on YouTube or your favorite podcast application so you don't miss a single episode. And do consider giving us a review and a nice positive rating. This podcast was brought to you by Tech Field Day, home of the IT experts from across the enterprise and a part of the Future Group.

com/podcast or view us on Techstrong tv. Thanks for listening and we'll see you next week.