 Hi, everyone. Thank you for joining us and welcome to Open InfraLive, the Open Infra Foundation's hour-long interactive show, sharing production use cases, open-source demos, industry conversations, and the latest updates from the global open infrastructure community. We're live here most Thursdays at 1,500 UTC streaming on YouTube and LinkedIn. My name is Jimmy MacArthur from the Open Infra Foundation, and I'll be your host for the day. As mentioned, we're streaming live and we'll be saving some time at the end of the episode for Q&A, so please drop your questions in the comments section throughout the show and we'll answer as many as we can. Our topic today is private clouds versus hyperscalers, a cost savings analysis. I'd like to welcome our guest, Todd Robinson, President of OpenMetal. Together, we'll examine the impact that cost has on your Cloud infrastructure choices. What are the advantages of going with private versus hyperscalers and how does open-source play a role in that choice? Todd, it's a pleasure to be with you. I'd like to start by noting this is great timing. We just met last week with Timothy Prickett Morgan to discuss his article, the on-premises empire strikes back at AWS from an open-source perspective, and it seems like the whole industry is buzzing about cost and repatriation. But before we get into that, I'm hoping you can give us a quick introduction into OpenMetal and the services that you provide. Hi, Jimmy. Thank you very much for having us today. As you mentioned, I'm with OpenMetal, and I've got a long history in the open-source world. I always like to share first that I'm open-source first and I am a profit second, and that to me is always an important thing, I think, being in this particular industry. I come from the era of MSQL. Sometimes I mistake it and get to the MySQL, but this is MSQL, this is a ProCGI days prior to many of the wonderful things that we have today. I'm with OpenMetal. OpenMetal is a passion project first originally spun out of our mother organization in motion hosting, and it is based upon our history and knowledge in open-source technologies, particularly OpenStack and Ceph, and it's been something that we've used for automation of our data centers, for providing services within the back end of InMotion and many of our other brands. Also, RamNote is one of our other brands where we use OpenStack and Ceph. We have all this history and institutional knowledge of how to use it for ourselves. We said, one of the things that when we think of open-source, and a concern that I've always had is these systems get more and more complicated. OpenStack is one of them, we love it but it is complicated, and things like CloudStack or Kafka or many of the Apache projects that are out there, they become so complicated that often it becomes very difficult for people who want to contribute, but maybe they're an individual to get access and to be able to meaningfully contribute to these company projects. We said, hey, you know what? With OpenStack and Ceph, we're already doing this for our companies internally. We can hyper-converge this, turn it into a production ready Cloud on top of three servers, and allow people access to a running RefStack validated OpenStack, and so that they can experience it and use it in a production environment with production ready, day two functionality included with it, to be able to let people run an OpenStack before having to architect an OpenStack. That's where we said, you know what? After we did that, we said, wow, you know what? This would be something that might really resonate with the industry and with the community. That's our history and I love talking about OpenSource and OpenStack, so I'm just happy to be here to cover those things with you. Yeah, absolutely. I love it. You're providing a way for, you don't have to be enterprise to enjoy OpenStack, and that's fantastic. I want to go to a blog post that you wrote recently talking about public versus private Cloud, and you've outlined your thoughts on it. In general, as I mentioned earlier, costs are top of mind for anybody that's running their workloads in the Cloud. Hyperscalers can offer a cost advantage at the onset because of free credits and incentives, but I'd like you to take us through the advantages of moving to private Cloud. How can you offset those advantages that the hyperscalers offer? Yeah, so I think I always like to talk about tipping points. I think we're going to get to that in a minute. That is a fundamental. There clearly are. OpenMetal, we're a combination really of bare metal. We do that. We absolutely do public Cloud, we absolutely do private Cloud, and we deliver private Cloud on-demand, which makes us something in between that. Our perspective always is that there are advantages to all the systems, and so you do want to take into account as you're working with this stuff, that there are times when those advantages are more important to you based upon your scale, based upon the type of work that you do, maybe based on your philosophy, maybe based on your region. There are certainly regions that are much more friendly to the private Cloud. The thing that we've got actually up in front of you is why OpenMetal, here at OpenMetal for our large customers, we selected a private first resource management. Actually, if you don't mind flip to the next one, it's I think it'll be easier for people to see. This is a graphic that we had, we're often talking about, okay, why did we make this choice? Why did we say, hey, for that we're going to take a private first versus the same approach that most of the public companies do, and it really comes down to a fundamental about the tenant. The tenant is our customer, the tenant that is working inside of the VM. When you're on a public Cloud, what's happening is what's up on the board. You can see the blue there is your average used resources, and this is something like if you were to look at your graph of your utilization, let's say CPU utilization over a day, that's going to be like a baseline. Then constantly, obviously, your workload is spiking, the CPU and the RAM and the IO is all being used up, but typically in a very spiky pattern above that. The hash mark blue takes into account that activity. In some cases, it spikes all the way up to the top, and in the red area, this is when you have to have inside of your virtual machine, you've got to have these resources available to accommodate for the spikes needing to go all the way up and max out whatever the resource might be, CPU, IO, etc. But in what actually happens in the public Cloud is your blue and your blue hash marks there, those are combined over time, and those essentially are your total average utilization. The red is what's wasted, and the point that we always try to make and why we ended up choosing a private first for our customers, is because in this scenario, the red use resources, they go back to the public Cloud. If you're running on a public Cloud, and the public Clouds for various approaches that they use, typically, they're going to resell that. That goes back to another customer, they're going to sell that again. In our case, when we said, hey, look, that doesn't philosophically, doesn't really match up with how we view our customer. Our customer should have access to that resources. They shouldn't have to be trying to plan their VMs really carefully in order to maximize the use of them, because most system administrators dev leaders will tell you, hey, don't try to put mixed workloads inside of servers or VMs in this case. Instead, you want to use VMs for what they are, keep them logically separated. Inside of this, that's one of the powers of private Cloud is, those unused resources, your private Cloud will simply return them to your other VMs. They will return them to the Cloud to distribute out, OpenStack to distribute out to your other VMs. That to me then, as it gives, it lets companies, it empowers companies to make the choice to say, I'm going to keep my workloads separate. I'm going to give them plenty of resources because I'm not paying on an individual resource basis, I'm paying for the aggregate. That allows companies to delegate down to their team to spin up new virtual private Cloud projects in the OpenStack vernacular, to spin up these projects and use them without having to like, oh, is that Nicola or Diane going to get us? Do I have to go to my manager to figure out if I can afford that or if I can do that? And so yeah, with this, the point that we were always trying to make this is that you can leverage the private Cloud in a way that lets your team move faster because they're not concerned about, like I've got to get everything that I possibly can get out of this VM. You don't have to, the OpenStack will handle that for you and will allow you to over-provision if you need to because that is now a shared resource and the Cloud is wonderful, of course, at being able to distribute that resources back and forth. So yeah, sorry, I ran on there a little bit long. What do you... Do you have a question on there? Yeah. No, I mean, I love it as a differentiator. It's obviously having a little bit more control over your VMs and exactly, I think where your dollar is going is important for all these companies, right? Yeah, absolutely. Yeah, absolutely. Yeah, this one for sure is it always just takes a bit, our attention always with these is again, we provide all these different services. So we're fans of public Cloud. We're fans of bare metal use cases. You know, we're fans of private Cloud, but it is always good to help educate people to know, like, look, this is what's really going on behind the scenes. And there are times when you're like, yeah, that's great. I don't care about that. I want to be spinning up my VMs on my public Cloud or maybe I'm distributed. So I need to have resources in hundreds of locations, which some of these cases have that. But I always like explaining to people the fundamentals that are underneath it. So as they learn, they're able to say, okay, I understand how that's playing in. So I get to make a decision about my Cloud, about how I want my resources to be distributed. Yeah, so this, yeah, this, sorry, go ahead. Yeah, I was gonna say, I love your philosophy. I mean, you mentioned earlier, open source first, profit second, which is, you know, but not the norm, I think, in the business. But I also love the philosophy of keeping the customer first and selling the customer's own paid resources back to other people in public Cloud is obviously not ideal, right? And what you're giving here is sort of a level playing field. And I think that's smart. I like it. Yeah, and I'll also say, so sometimes when you're buying from the public Cloud, you are buying more fixed resources. And so they may not be doing this. Definitely, there's many cases where this is happening, but you also may be buying like kind of almost like a bare metal, like, and that you're assuredly getting those resources. The issue with it is, is you still, you still are not gonna wanna cram absolutely everything you can into that VM to actually get your full value out of it, to actually use the resource. You're not gonna wanna do that because yeah, then you're gonna end up, you know, mixing things that shouldn't be mixed or you're maxing out things that in opportune times. And so I always say, yeah, one of the powers of public Cloud or private Cloud is that you have this level of control. You decide where those extra resources go and when they go there. Yeah, that's fantastic. I wanted to touch on one other thing. And on your blog post, you mentioned David Hansen, the Ruby on Rails creator who had a very public breakup with a hyperscaler public Cloud. Any similar stories from your perspective, from open metals perspective of customers that had been sort of fed up with the hyperscalers and moved on to your services? Yeah, and I've got a little, just because I followed the provider, they're 37 signals, right? They do base camp and hay. And so they had the advantage of running actually hybrid, I guess could be the term is, so they had always run base camp inside of a data center and they were running hay out on the public Cloud. And so they had a lot of the skill sets that are necessary and the institutional knowledge themselves to help them go like, well, wait a second, like why is this public Cloud so expensive? Cause they have a direct comparison. And so they were able to make choices pretty easily to say like, wait a second, this just doesn't make sense. Plus I already have these skill sets. And yeah, so yeah, I followed them many times over the years as a base camp user early on in my career. And so we find that unless you have that perspective where you actually are seeing both of them literally at the same time, and so that you can very clearly make that choice, you have to get out there and explore what might be your other options. Typically you're gonna do this when the numbers get pretty significant. And so they were in numbers that were very significant, like they were very public about posting all these things. And so at $10,000 a month for a public Cloud, does that make sense for you to start looking? Maybe at 20 or at 30 or at 40. And they were starting to get into seven of these numbers that were very large and they were easily able to compare it. Yeah, and so our case, one of the public company, companies that has been public about their usage of us is a company called Convezio. It's a really cool WordPress hostings like software as a service approach. And they themselves were facing that same thing as very large expenses associated with things that they would think of as fundamental to the performance of their product for their customer. And having to pay like a really large expense for those things. And then having to like end up with a question of like, do I choose the most powerful best thing for my customer where it's an expense that's exaggerated against what their business model could support, right? But their side, their philosophy was customer first, which of course we love is customer first. So they were really driven to the decision to say, like, no, I have to go out and find out what are the other options that are out there because I have to have that performance. Boy, I can't do it at that cost. And some of my competitors are out there doing it, but they're skimping a little bit on the performance. And so they were driven to start shopping around for performance first and then costs second, which we love because if you tune your private cloud properly, you're gonna get that performance and you're gonna get the cost savings. Yeah, so Convezio is a great example to check them out of their customer verse. We love that. Pipestream is another company that was, they use this for their development pipelines. And they had found development pipelines are kind of an interesting workload because when the team is working and they're pushing, there's lots of activity and they need a lot of resources, but typically you also don't wanna tear it down when they're not working. And tearing it down requires its own set of work in order to create pipelines like that. And so they ended up moving over to us because they were looking at the expense that they had in their pipeline and saying like, this is just for the amount that we actually use, why is this thing so expensive? And us being, again, OpenStack is cloud native, call the API, the API is wonderful. It does exactly what you would expect it to do just like any other cloud provider. And yeah, so Pipestream is another example. So typically for us, we're seeing softwares or service companies, we're seeing OpenStack companies, companies that are like, look, I've been running my own OpenStack, but you guys do it faster and easier for us. We're seeing that come over to us. Yeah, did that give you an idea there, Jimmy? Yeah, no, that's fantastic. And I think you bring up a good point, which is the scale, right? If I'm a person that's starting my business, probably hyperscalers is just fine to get started. Obviously we prefer OpenStack powered public clouds of which there are many and some fine choices out there. But it brings to mind something that you mentioned in another blog post, which is the cost tipping point, right? So like at what point do you scale so much that obviously 37 signals in base camp or, hey, those are extreme examples of like very, very high traffic websites, but you know, order services and for the regular Joe, who's starting up a business, what's the point that you start to get to to really make the move? Yeah, so we see, yeah, the tipping point is really is a real thing that is wise to be aware of and to also note that tipping point, especially in the private area, has been increasingly coming down. And so as automation has occurred in the private, let's call it the private world, it's data center automation, OpenStack's ironic or by frost, we use it, it's a wonderful automation system. Those improvements that have been going on in that world has lowered the tipping point, but tipping points are typically made up of a couple of different things. So of course it's gonna be your compute, how much compute are you realistically using? What's your egress? Egress is a pretty expensive item on the public cloud because they have found that as a relatively straightforward place to make a pretty big margin. And so if you're egress, you're gonna be looking at egress and if your egress is very expensive in the private world, let's call it if you're running, if you're either out of your own data center, which I think is less and less common for the mid-sized company, but if you're running out of a co-location facility or if you could be running out of a co-location facility, certainly you do gotta take into account like, hey, look, I might have to have my network devices, I've gotta have my connectivity, but in a lot of cases, you can be buying some of that trivially from your data center provider, your co-loc provider. Or in our case, it's the same thing, is where that blend of between the public and the private. And so yeah, egress is a common tipping point just because it can really add up. We've seen bills, especially for companies that are moving hundreds of terabytes and you start getting up into the petabytes, you can be in 30, 40, $50,000 a month on those things. And I think what you've got up on the board right there is just a quick analysis of what the public cloud can end up running you. And so yeah, I'll just pick on the third one down there. So we call it the large deployments, 1,000 VMs, 150 terabytes. Those are relatively small VMs in this comparison. So if you're using very large VMs, like 64 BCPU type ones, that's a much smaller number that might only be like a hundred in this case. But what you can see on the public cloud is, and we use the 150 terabytes of egress and keeping in mind, they're often charging like between four and a half cents, five, five and a half cents per gig for this. Those numbers really add up significantly. As egress definitely is one of them where you'll find people go like, hey, I've got to figure out a different solution for this. The VMs themselves, depending on what you're choosing from the catalog that might be available, there are ways to be more economical on the public cloud. But typically, anything that is offered on the public cloud, you're gonna find on the private cloud, either by providers like us or many of the other ones that are offering it out there. And so, yeah, and so what you'll see, and I think on the board there is, for example, that third level one, your yearly difference might be coming in at 170-ish to 180-ish thousand dollars. And so those are ones where you start to say, hey, in our case, what's on the board there is we're comparing it with a managed private cloud. And so that's oftentimes easier for companies that are coming off of the public cloud to not have to step fully into the complete management of the system. And so we will often not do comparisons against the managed private cloud. So that's when you've got a company that you contracted with that helps you run your private cloud. Over time, in our case, we encourage our customers to learn more and more about it. And as they become more comfortable with the automation that we've used for managing the cloud, they have a choice that they can make a choice to shift in between managing it themselves or having us or a third party manager. But as you can see, the numbers really can add up there. It isn't for public cloud for smaller installations are wonderful. It's exactly what it's for. And it's easily a more economical thing to do. So definitely like you can see the very first one, there are small 100 VMs, 10 terabytes of transfer. If you're really looking at saving 18,000 or so a year, you're probably in the spot where maybe philosophically you don't wanna be on the public cloud. And so it's a reasonable thing to make a change over to a private cloud at that stage. But typically that wouldn't be a cost-driven decision. But yeah, as you go up in these levels, again, watch out for your egress, watch out for your storage costs. Let's just use the typical S3. Certainly you can be less expensive. But one of the things we also find is that companies are looking, I'll proffer this as a category, is speed of completing a job. And this just means like if you're doing a modeling job, for example, and you need your compute to finish up with its model and it needs to push its data off into an S3 bucket and you need to, and then once that's done, you're gonna pick up another job. Well, what we found is that by physical location, it actually really comes into play in the speed of your job. And so that you can do your compute, get that model done, shove it off into an S3 bucket that's literally like next door, right? Network-wise and complete your workload in a more timely fashion. So there's also some of these hidden tipping points that have to do with like looking at how long does it actually take me to process this stuff? And it is moving data around is a big part of that. Okay, I'll head on for a bit there too, Jamie. What other, yeah, what questions are there? No, you're great. I mean, I think one of the common arguments against private cloud is that like once you factor in all the costs of operations and data center costs and procurement, et cetera, that in the end, all the cloud is less expensive. And I think this is a good way to look at it to understand that there are costs that are outside or sort of separate from what your regular bill looks like. So yeah, this is good stuff. It definitely is, the tipping points often come up and it typically is gonna come up for a manager or a leader in the IT side. It may be driven by the CFO, the CFO going like, wow, these numbers are large or I'm getting all this spikiness. And so we're now out of budget and what's going on. And so that's often one of the drivers. And then I would just encourage those leaders that do end up facing this question of like, how do I need to drive this down? Is there are lots of places to go and you do want to challenge this thought that the public cloud is less expensive. Because in many cases it's not, it is a tipping point. And there are a lot of options. And the private area, whether it's that you're gonna decide to run your own co-location facility, get your own hardware, or you're gonna use automated bare metal from a provider. But I would just encourage you, like there are a lot of options out there that have had significant improvements over the last years. So as public cloud was kind of blowing up and everybody's like, oh, everybody's gotta get onto the public cloud. There was a lot of things going on for the private area. And so it's a lot more efficient than it was 10 or 15 years ago. And I would just encourage people to open up again to the concept of that you can run a lot of this stuff yourself or you can run it with partners. And it's still cloud native. You're still doing the same work. You're still using APIs, just like your team has probably become used to. Yeah, it's funny because I think that everybody thinks of moving to the cloud as this, you know, there's the one approach and you get locked in and we can't run on-prem anymore and this and that. And you're exactly right. There are plenty of options out there. There's plenty of things. And you said something when we were talking earlier about it's not a move to the cloud. It's not a physical destination, but it's an approach. I'd love to hear sort of a little bit of expounding on that philosophy. Oh yeah, absolutely. And I would say I've heard this terminology from a few people. I wish I knew exactly where, when someone had said this to me, like cloud is not a physical destination. It is a philosophy. You don't move, like get up and move to the cloud. It's adopting philosophies like API first. It's adopting philosophies of being able to deploy infrastructure in a automated, trivial fashion. Infrastructure is code. These are, for me, that is the definition of cloud. Not a physical location, but the hyperscalers have definitely kind of got that set in people's mind like I need to move to the cloud. Well, in fact, it's techniques and approaches versus actually a physical move. So when you go back and forth between providers, that might be a physical move, but that's a provider. That is not a philosophy. And I would definitely encourage people, both our company has made this transition over the many years like our mother company started in like 2001. So there wasn't this cloud philosophy, right? There wasn't this API separation, let one service do its thing, let the other service do its thing and those could be independently developed. Instead, we had to make that transition. And so we absolutely have some of the old systems as well as the new systems, but we didn't move. We just learned that, hey, that is a more efficient way to do it and let free up the developers to be able to spawn up resources as they see fit. But that's open stack is, I would say, like a lot of times when I see services out there and I go, oh, like Barbican and then I see like HashiCorp's vault, and I'm like, wait a minute, boy, that really looks like Barbican. What is that exactly? Yeah, yeah, right. And then you can say like, there really is a lot of what the cloud is came out of what open stack was doing. And the philosophy, it's not necessarily the exact software systems or API and how the APIs work, but the philosophy and I would say open stack was a real leader and continues to push that. And but I would still go back to like, I think it's healthy for people to think of like, when you move back and forth between good providers that are modern, they're all gonna be API first. This is the cloud, right? And you can use that to your advantage knowing that, no, I'm not gonna be asking my team to like re-engineer a whole bunch of stuff. Instead, I'm going to say, does the Terraform provider have the functionality that I need? Oh, yes it does, look at that. And you're gonna find, of course, a lot of people are running in Terraform and you can move that back and forth between the cloud providers. But so yeah, cloud is not a physical destination. It's a philosophical destination of making using resources more easily and tricky. Yeah, that's fantastic. I think there's been, people talk a lot about hybrid cloud. I think it's a term that's confusing to folks, but kind of the way you're laying it out makes a lot of sense. It's just a choice of where best to put your workloads and what makes the most sense from an efficiency perspective. And I think that's great. Speaking of efficiency. So we've talked a lot about cost efficiency, but another thing that's kind of top of mind for consumers, for the world, I suppose, is resource consumption, power consumption and how that affects our day-to-day lives. Sustainability is really something that there's repatriation, which seems to be the word of 2023. And then there's sustainability, which I think is just this ongoing challenge in the industry. How is open metal tackling that and how do you see OpenStack playing a role in that moving forward? Yeah, that's an interesting question. So definitely on the repatriation, I would say this, watch out for the nomenclature. When you repatriate, if you're gonna be moving workloads off of the mega clouds or the hyperscalers, hopefully you're, you've already targeted OpenStack as the place that you're gonna be bringing them into. And so in that case, you're still on the cloud, as you get to still use the API. So repatriation is really just moving providers. And maybe you're gonna be doing it yourself. Maybe you're gonna be doing it on bare metal, bare metal provider. Maybe you're gonna be doing it on somebody like us. So yeah, repatriation, I think it's a great concept. It really just is moving another, moving to another provider, you're still gonna be in the cloud, you're still gonna be doing that. For sustainability and looking forward, I think for me is I've always, this is, there's several things that we do, but I'll focus on this one is that you as we, as we provision and use these large workload systems, whether they're servers or whether you're on a VM somewhere, is that you do wanna be wise about what you're doing. You wanna understand what's going on underneath if you really want to have a kind of a whole picture of what your utilization might be. And I would definitely encourage people like, we run into this even with there's, sometimes there's a philosophy of like, hold on to every piece of data that you ever had. And I would just say I would encourage people like that, doesn't ever end. So if you have that philosophy, right, it doesn't end and there's several things. So one, yeah, it's gonna cost you, it's gonna cost you dollars to keep that. But it also is it has a broader reach is like, if you're buying those servers or you're buying that space, somebody is buying a server, like people go, hey, serverless, serverless is top really effect. There's of course a server and you are utilizing those resources and those resources means somebody has built that piece of that device. They have pulled the rare earth or whatever out of the ground they have turned that into a server, they put it on a ship, they got it into a data center, they're plugging it in. So I just say for sure in our space, that is absolutely the path that we're going on, right? Like as people do need these services, I would just say be wise about those things, understand that both your cost is dependent on decisions that you're making that might also have impact on resource usage that has an impact across the world. So yeah, I'm not sure, Jimmy, if that's I'm gonna be able to answer as many of those questions. I love things about like our data center in Amsterdam. My understanding does not have this, but like for example, there are data centers in Amsterdam that the hot water, the cooling water is delivered to the city to use as hot water for the city. So those are the kinds of things like that. When I hear people come up with these kind of like innovative use cases and I go like, that's pretty clever. I always like to just shine lights on things like that as people start to make changes to say, well, wait a second, let's look at this a little bit differently. You know, data centers are here to stay. I mean, yes, they absolutely are, but there are ways to have them serve multi-purpose. So that was like a great invention. So yeah, I can't speak enough on it, I think, but hopefully that gives you some perspective. Yeah, absolutely. Yeah, I think we've got a couple of those folks in our community, Cloud and Heat transfers their, the heat generated from the data center back into apartment buildings and the cloud does something similar. Very good. But yeah, I actually, I wanted to ask, I know that you have data center in the Netherlands and then obviously in the U.S. as well. Part of, I think people moving back to, you know, open sac powered public clouds and open sac powered private clouds is because of this kind of push by the European Union for GDPR and kind of data sovereignty. At the same time, that doesn't seem to be such a concern for folks in the U.S., but I wonder, do you see that trend changing? Do you see U.S. companies and U.S. citizens kind of rethinking their own sort of privacy implications? I think the U.S., there is pretty definitive differences I think between our European and our U.S. base. The Europeans, I think there's many factors actually that have the Europeans approaching how they do these things differently. One of them is, is that the hyperscalers, these are American companies, that, you know, for lack of a better term are coming in on the European turf without necessarily having the same philosophical alignment as many of the Europeans do. So first off is that that's an impact is that I think here in the U.S., they're American companies, we're proud of those companies in many cases, we've grown up with them a bit and I think our view of the brands tend to be more positive than the Europeans. So first off, the Europeans tend, I think are very much looking at those organizations and saying first off, they don't align philosophically and then second off, they're not from here. And so it's a little bit, it's a lot more difficult to deal with them. Now obviously they have introduced many laws to help those companies, encourage those companies, let's say, to align with their philosophies. So that we noticed first is in some cases, we have a wonderful company, my mini factory who is with us and they just, they're like, yeah, we don't wanna work with them. Like that's just something that's important to us. We don't want to work with them versus working with a company that is very, they're a very open source, a very grassroots company. And so yeah, I think what we see the difference I think you're right in that the US some companies don't think as much about data sovereignty because they are just kind of used to, hey, this is what I work under and I know where my data is, my data is here. So yeah, I would say there is quite a bit of a difference. I don't, do I see, you'd ask like, do we see if the US companies are taking it more serious? Are they're looking at it more like? I guess what I'm, what I'm seeing is more, for instance, there's an organization called O-Rock that is private cloud specifically built for the government to government standards but that's running an open stack. I'm seeing more open stack powered public clouds pop up which three, four years ago, we wouldn't have seen that. As we, you and I were talking about before the show, open metal might have seemed like a risk three or four years ago to a lot of folks but you've hit this niche where it's, there is a turning point and I guess in the US it's cost, right? And in Europe it's a combination of cost and data sovereignty kind of driving that but I do feel like people are becoming a little bit more cognizant of it here in the States and I don't know, maybe I'm crazy though. I think first we, the first thing that's on people's minds is cost. I think that that's the world that we're in and yeah, you had mentioned, did we face an uphill battle when we first said, one of my closest friends in the world said to me, he was like, what are you doing? Like why would you try to take on Amazon or somebody like that? Why would you do that? Now this was a few years back and he says that's just foolish, you know? Like going up against a company, those sets of companies like that. And in some ways, yeah, he was probably right, there's some foolishness inside of there but also I would say like, there is a place for this type of product, for open stack to be vibrant at the mid-level. They were, as an enterprise, of course, it was always a logical choice and for data center managers, companies like us who had to run each and it was always a logical choice. It's a brilliant piece of software, our brilliant set of software. But to bring it to the mid-level, yeah, I think that what is driving people though to consider it here in the US mostly is cost, is they've hit a tipping point and they make decisions like, when those numbers get big like that, 20 or 30 or $40,000 a month that you're saving, when they realize that, look, I can take those dollars and I can hire more developers and I can move the product faster. It's not often like, yeah, I'm just gonna drive that to the bottom line, but there's definitely some companies that are obviously gonna do that. But some of it is like, no, it helps me move faster because I'm not blowing all that money on this part of my business. Instead, I get to put it towards my product. A lot of the SaaS companies have that. They're gonna come to us and that's what they're saying. First is like, I just wanna move the dollars out of here, over here, because this is valuable to me. But do we see them as data sovereignty? I would say OROC and we definitely have companies that do, especially when working for government agencies and things like that, that very much do follow that. To be honest, it's really two things. It's a control issue with performance. So they're like, look, my ratios, I need lots of RAM or I need a little bit of RAM or I need lots of IO and none of the flavors that I can buy from the public clouds are like that. And so I'm ending up having to buy something that has tons of NVMEs attached to it just because I need a lot of VCPU or I need a lot of CPU. And so it's just, it's kind of a waste of money. So you see that often it's a performance discussion, but cost is very much top line in the US. The Europeans I think are very much more aligned with what you're saying, like that I wanna understand where my data is, where my customers are the visitors or users of my application, where their information is gonna be, they wanna be more empowered. I think that that's also a piece of that. But I would say though that it feels like the US market like a fast forward a couple of years from now. And as people learn that this type of product is available and that OpenStack is now more accessible and OpenStack has become easier. I would, for everybody listening, I think you already know this, but OpenStack has become easier and easier to use. There are a lot of these small systems that allow you to fire it up on a single computer and follow through on getting familiar and using it. But I think fast forward a few years from now and because this ease of use has become more prevalent or the market has become more aware, I think the empowerment part will become more clear as people will come in and say, I want that because I want access to it. I wanna be able to modify my flavors so that it's as fast as it can possibly be because I need my workflow to get done or my customers are doing something and I want it to be fast for them. Yeah, no, that makes a lot of sense I think. We've got a question from the audience. So sometimes it can be hard for folks to go all in on private cloud. So we see a lot of users turning to hybrid. When doing this, what types of workloads are best for private cloud over public cloud? Yeah, I definitely would say going all in on private cloud, you would do that over time. So I think that's what you're getting at is is that it has to do first with understanding what its capabilities are. And then, so yeah, specifically if I was looking at workloads, so some of them that we see are big math models, might be someplace where you're like, because I need a ton of access to CPU but I'm gonna be shipping the data out. But you go out to the public cloud, you can't buy a flavor that's just all CPU and RAM and basically no disk. You're gonna buy something that's gonna come with a bunch of, probably a 30 year cost is gonna be something that you don't need. So that could be an area where you're coming in. This flavor, what I was talking about about flavors is when you have workloads that don't fit those common ratio of VM types flavors that you can buy out there. Your tip, look at your tipping points. And so the private cloud, I think when people say that sometimes egress is like an easy one, is if you've got really large egress bills, you should start shopping. You should start looking around to figure out what are some of the alternatives that are out there. And that's a funny one because that's not really private cloud. That doesn't have to do with private cloud. That has to do with how you purchase your bandwidth from a provider. But because you need to stay cloud native or I would advise that you stay cloud native, what you are gonna find is that you may end up on a private cloud because you do need to make the transition to essentially either buying your own bandwidth from a provider or going with a provider that's gonna give you a price that's logical for you. So definitely it's gonna be, you're looking for your tipping points. You're gonna look at your workloads. Find the most expensive ones. Start with that. So you can always say like, all right, let me open up this big bill of mine, which it can be a bit tricky, but you open up the bill and you gotta start looking for commonalities. And so yeah, if you've got let's say 50 big VMs, the really large ones and you're paying $3,000 a month or something for these really large VMs, but you've got commonality there, you can spot that. You'll be able to see those in your bills and you say, okay, that's an area where I should focus. So yeah, start with that. Start with the bill, take a look through to find commonalities and workloads. And those would be ones that I would often target. And I would definitely encourage now, this is me as open metal talking, of course, I would encourage you to, before you go all in and private cloud is spend some time with some of the organizations that will help you through that. There's great contractors out there that can say, look, I can help you build that cloud and I can do it in a way that's not oversized. Remember like OpenStack and SEP, they're designed around adding and removing resources within there. So don't go into this like thinking like, I need to build this huge cloud, you don't need to do that. These systems are fantastic at being able to add and remove resources to expand and contract your cloud. So I would say try not to go with a, I need to expect this for a really large, start with a small one and understand how those things work, but they are scalable. Those systems are very scalable. That's one of the improvements over these many previous years that you can rely on. They're very good at adding and removing resources. Yeah, and an advantage, that's the cloud advantage, right? It's, you don't have to go on-prem, you don't have to plan out a whole data center. You've already got that available and you can pick and choose and see what fits right. Yep, you got it. Love it. Well, I think, I don't know if there's anything you want to close with, Todd. Yeah, I think I would always like to kind of go back to this concept of stepping into the cloud, stepping into the private world or where infrastructure, you're going to be a little bit closer to infrastructure and recommend to people whenever you're looking through choices that you have that people are often fearful and they say, well, I'm going to have to run the hardware again or I'm going to have to do procurement again. But look around, there's areas that you can allow your other providers to do those things for you. But that you can also look within your team to recognize that many times running servers didn't change when you moved into VMs. This is actually one of the points that the 37 signals was making that David was making, is that when you run VMs, you have system administration skills. Your DevOps team has skills that are, they know how to run their stacks. They know how to run operating systems. They know these things. And if they have stepped away from a little bit and companies over the years by moving to the private public cloud, sometimes the message was, hey, you're not going to have to have system administrators anymore. And reality is, no, you're still running servers. And in some cases, you're running hundreds or thousands of servers still, you have to have a lot of really good skills. Now, what I could say is, and 37 signals experienced this, is they didn't have additive staff when they brought Hay in-house or as they're bringing Hay in-house today, because they already had a skillset of system administrators that were running inside of VMs. And so those skillsets still exist and you still need to have them. And in fact, for a lot of software as a service companies, that's going to be one of your advantages is how effective is your stack inside of a VM or on bare metal or wherever it happens to be in a container or something like that. And so I would say is like that there is a lot, typically there is quite a bit of skillset still inside of a company that is going to be running servers. And so you don't have to be so afraid of some of these running the bare metal itself because in this case, a lot of your team has that. And you can also look at as an opportunity for the team to get even better, the team that's managing your VM to even get even better at the stack that they're running, right? And it gives them advantage when they get a little more fundamental to say, wow, actually, I know how to carry this back up to the customer like that. And I can make my whole system faster because now I have a better fundamental understanding. So yeah, so I would just end with that is like, I know sometimes it can be a little bit nerve wracking to say, well, we don't have those skillsets and so how are we going to take on this extra work? But in many cases, you have staff that are skilled at your servers anyways, they just happen to be inside of VMs. They can run bare metal just as well. Love it. Well, listen, we are about out of time here and I wanna thank you very much for taking some time out of your day and talking with our audience today. I really appreciated the conversation. Yeah, Jim, you know, thank you for having me. And again, I'm just excited that the industry is now talking so much about this and that more and more options are now becoming or light is being shined on more and more options and that OpenStack is front and center and OpenStack has been growing dramatically in these last couple of years, rightfully so because it has this capability and it stayed true to its API first nature. And so like to me, I'm just excited for OpenStack. I'll give a shout out to Seth as well. Seth is a fan, I'm a fan, it's a fan favorite for me when I was around the time when Seth had spun out a dream host. And so I'm also excited to see it taking such a major role in here alongside of OpenStack. But yeah, just very excited and I know I think we've got an upcoming meet and greet but yeah, I don't know if you wanted to shift over to that and talk a little bit about the upcoming conference here, Jimmy, before we cut. Yeah, absolutely. First, let me say for those of you that have DevOps folks that are considering moving to private cloud, OpenMetal has some fantastic tutorials on how to jump into OpenStack. Check out their LinkedIn page. I don't think I have a link ready but I wanted to plug that and say thanks to OpenMetal for really supporting the OpenStack project. We greatly appreciate it. And yes, as Todd said, join us in Vancouver, June 13th to 15th, OpenMetal will be there. They are a sponsor of the event and so will many of our other organizations. And we'd like to thank sponsors of the event, our headline sponsor, of course, Wind River and our premier sponsor, Ocastro and all of our exhibitor sponsors, including OpenMetal. Thanks so much for supporting the event and if you have any interest in joining the event or sponsoring it, hit me up at jimmyatopenim.dev and I'm happy to provide some more information on that. And remember if you have an idea for a future episode, submit your ideas at ideas.openinfra.live and perhaps we'll see you on a future show. And finally, I'd like to thank all of our Open Infra Foundation members to make all of this possible. If your organization would like to join the Open Infra Foundation, take a look at openinfra.dev.join. Thanks much and have a wonderful day.