Download Otter for your meeting notes

20250401 Lab vs Enterprise: Otter Voice Meeting Notes

20250401 Lab vs Enterprise

Rob Hirschfeld37min

Rob Hirschfeld

00:06

Rob, Hello, I'm Rob Hirschfeld, CEO and co founder of RackN and your host for the cloud 2030 podcast in this episode of the DevOps Lunch and Learn we talk about home Loving versus enterprise use cases and why it is so tricky to satisfy the home user with enterprise products, really is one of those dilemmas, because we love to see more crossover, and it just really is a struggle, and we're going to Talk about why. I know you'll enjoy the podcast.

Rob Hirschfeld

00:47

I actually would turn us to the topic of the day if you all are interested, because this was, this was curious, and we set it up two weeks ago. It's this, and it's related to what we're just saying. It's this divide between home labs and enterprise use cases, and I can give a little bit so the questions I had as reminders is, why is it so hard to translate home use into the enterprise use cases? And why is it so hard for enterprise products to translate into home use. And this is near and dear to, you know, our heart at RackN, because we're we always want to help people be like, Oh, you could use us in your home lab. And you know, you get all the these cool benefits, but it's been really hard for us to find that as a useful, you know, be useful. And then when we do help people in their home labs, the stuff that they need doesn't translate back into anything else we do. And so it felt, to me, to be sort of a universal truth in this. But I wanted to explore it right. How can, can we bridge the divide is, you know,

Unknown Speaker

02:05

is the divide real? It's

Speaker 1

02:07

definitely real. It is actually something I have spent the last handful of weeks working on, as I need to get VMware out of my home lab at some point in the near future before my license runs out, and as part of my looking into solutions in the market, I ultimately landed on the moment to say, crazy endeavor of quote, unquote, building my own hypervisor. Now, okay, it's a very loose term. Essentially, all it really is is taking KVM, live, vert, Q, mu, open, V switch and building a wrapper around it. But the reason for me justifying it, obviously it's a learning experience to continue to get deeper into that. But also, the other thing for me was thinking about the common thing that people will ask from a small company that's just starting out is, will your solution scale up to the enterprise, or scale up to 1000s of workloads, or whatever it might be, there is a argument that can be made there about that capability. To me, the inverse can also be made. A solution built to handle 1000s of VMs, or enterprise scale or global scale is not necessarily built well to scale down in that I don't need all the knobs and dials. And one would say that was the common argument against open stack several years ago, or in the earlier days, was it's too complex and too cumbersome for me to want to deploy it in my home lab, because I'm spending half of my time trying to wire everything together. I would really like something that is extremely simple and bare bones to get started, and HashiCorp actually does a pretty good job of this in that there's on things like console and vault and some of the others, there's a dev mode in that the goal is literally just to get you up and running in one simple command cuts out a lot of the additional complexity and interconnection between different components, and lets you really start to see that experience.

Claus Strommer

04:05

That's the home lab argument of COVID versus Docker, essentially,

Rob Hirschfeld

04:16

of of wanting to have a simple container scheduler versus a complex container scheduler? Yeah,

Claus Strommer

04:21

yeah. If you, if all you you're doing is just starting on your home lab or just maintaining it on a single server, do you really need COVID? Like, if you don't hear what the source is, automatically restarting or migrating things, or balancing the law between multiple nodes, just like only, all you want to do is just run a container, just pronounce this oracle for you. In that scenario, there's

Rob Hirschfeld

04:53

a lot, yeah. So do you see something like K, 3s where some of that complexity, um. Was taken out, like all the SED pieces eliminated, the distributed host. Does that address the complexity? Is it? Is it just the complexity of the the scaffolding, or is it the APIs and the command line and the pods and all that? Where does the complexity hit? A

Claus Strommer

05:23

little bit of both the one before k3 is or mini cube, or micro face, or whatever. The biggest issue was you couldn't just install a cluster, let's say cube ADM, and be ready to go. You also need to set up your CNI. You needed to, you to set up your storage. You, and then if you didn't pay attention, you needed to rotate your search. And that was a pain back then. Yeah. Oh, my God, Joe, yeah. But even then they're like, going from I'm running a container and Docker to, okay. Now I can't just run the container in coordinates. I need to create a manifest, and I need to define all of these conditions are on it, like the deployment I need, the spec and so on. It is a cognitive challenge, particularly for people who are not familiar, who don't have a template to begin with, yeah? Like, if you asked me to create a deployment from scratch without looking at the reference, I wouldn't be able to. I would like, if I'm going to create a deployment, most likely I would just do I would create a new helm project and let that create a template for me and then start modifying it. But that's because I know the tooling, and I have a certain and I know that for me, how gives me certain consistency. I I know it's not for everyone. So some people like keep spray or just um, customized, customized, yes, yeah, but, but anyway, the bottom line is that hardly anyone curates manifest from scratch, unless you're starting to learn right and you and when you're a home lover, typically, you don't start from scratch. You start with what you already have, and then you're like, Okay, how can I translate this to goodness? On, on, there's a barrier there. It's, it's an understandable barrier. Once you realize everything the Coon is gives you, but if you don't need any everything that the Coronavirus gives you, you're going to come out disappointed, let's

Speaker 1

08:18

say, from a home lab perspective, particularly with Kubernetes networking is one primary one in that, if you're used to a hypervisor or VMs, typically it's going to drop it on the same l2 network. I can get to my VMs. I can get the things I need, or even from a doctor standpoint, as long as I'm exposing that port, utilizing the host exposing I can get to my web server that I just deployed via the Docker container, Kubernetes. It's more of a leap for most people, is it's the best way for me to expose a web server from my Kubernetes cluster. Obviously, you have solutions like metal LB and some of the others that, in essence, provide the l2 accessibility. You could also do via the in addition to that, the ingress controllers and those sorts of things. But there's no I just want to run this on my my flat network, for the most part, out of the box. In many cases, that's a challenge, I think, for most people, whether it's home lab or in an enterprise setting, is figuring out what to do about that the same in many ways, sorts of things apply to the storage side as well. Gotta use persistent volume claims and persistent volumes, and how do I map that to the underlying storage? There's a lot of new concepts and constructs to understand that, I think, make it challenging to spend a portion of your time tinkering on it in your home lab.

Claus Strommer

09:53

Yeah, I agree, like it's the it's the difference between you. Like that, which with COVID, is you basically have to take a course to get started. You have some you have to have someone guide you, like, whether it's whether it's a, like a self guided, like online course or physical course, or whatever, you have to follow up plan to start learning with, with VMs and, to a smaller degree, Docker, because Docker has its own networking issues. Yeah, I see Martinez point where you can just run things as you go, like you start a service. You figure out where listening to like, what address you hit, the address you're likely able to connect to it. As you start containerizing with Docker, that becomes a little bit more difficult, like, you need to know the difference between an exposed port and a map port. And then with Coronavirus, yes, yes, you have the concept of a service in front of that, which, unless you've already learned things like console, you're not familiar with that.

Rob Hirschfeld

11:14

So, oh my goodness, this idea, and maybe there's a divide between corporate so if my assumption has been, if you're building a home lab, it's, it's likely that you're an IT worker, and you've you're learning, you're doing it to learn something right. The home lab is not just, I want to store family photos and videos or or do something like I'm building I'm building something because it has work utility outside. But what you're describing is you're not going to learn Kubernetes, you know, doing a home lab install. This is that that's what am I hearing that right? You're basically court. You know it, you know it's

Speaker 1

12:05

gonna depend on the desire. So like, as an example, there's a lot of testing that I do where I need a platform, but I'm not testing the platform at that moment. I need to test, let's say, data protection for Kubernetes workloads. I don't care what cluster it is, per se, I don't necessarily care about a lot of the details of the cluster. I'm concerned about that component or I need to test an application that might run on Windows. I don't need to care about doing bunch of windows updates and all the things that come along with it. So I would say it depends on which piece that you're trying to test at a given moment in time, in terms of what infrastructure you're going to need.

Claus Strommer

12:43

Okay, I would also question that the premise that the home lovers are home ladders are largely IT professionals, because any, essentially, anyone who runs this service at home, a lot like an ass or a media center or whatever, they fall under the home library category, and A large portion of those, they really just want convenience and access to their to the media that they own, right?

Rob Hirschfeld

13:28

But I would expect those are going to be single box installs or they're going to be more pliancy pieces. They're not, you know, you're not, but you're not expecting that to become an enterprise solution. And you know, to Martez first point, all the bells and whistles are probably the antithesis of what you want.

Claus Strommer

13:48

Yeah, so, so the these, these people don't want all of the bells and whistles that I agree with that, but that, if you follow the home lover communities, you see plenty of cases where people outgrow their single appliance. Well, let's say that the start with a single like, let's say with with the hard drive attached to the computer that they moved to an off the shelf like to Bay NAS, and then five years later that they've outgrown it, and they look at the new like four or 8b net NAS devices, and they sweat at the price, and then they start wondering, Well, can I build this myself? Like all I need is access to my media, and they don't necessarily come from an IT background. And they don't necessarily want all the buzz and whistles, but they want they have certain expectations about convenience and how access to their content or their home services goes and they. They want to retain that, and it's not just media either. Is like home automation is another big one now with home assistant like the people that want the automation but don't want it tied to a cloud provider like Alexa or Google nest, so there's still a lot of the demand. But again, unless you're an IT professional, which, again, my argument is that significant portions are not then you don't have the know. How about things like port mapping and services and load balancing and so on right. And you're, you're going to want a solution that is closer to what you already know, like, kind of like, like the dash, right? Like, if, like, if all you have is a hammer, everything starts to look like a nail. Well,

Rob Hirschfeld

16:08

edge, the edge, to me, is a sort of a classic example of the of where this plays out, because feels like a lot of the edge solutions are very laby home lobby, in the way they, they, you know, they're like, Oh, we're going to use nooks, and we're going to, you know, do you know, assume USB installs and updates and, you know, they're not thinking remote management. You know, in what I see, they're, they're set up as huge, get, get ops environments.

Rocky

16:43

And yet, at the same time, all the corporations that are looking at the edge are looking at cloud connectivity. They're not looking at isolated or many mini grids, if you

Rob Hirschfeld

16:58

will. Right? No, the expertise is really hard to set up.

Rocky

17:02

Yeah, bifurcation going on there?

Rob Hirschfeld

17:06

Well, that's, you know, if you're dealing with a home lab, you are assuming you're there. You're going to fix it, you're going to log into the keyboard, right? There's not a lot of benefit to having a, you know, and you know, sign I if you're an IT professional, you might be different, but Right, doing something where you're automatically doing configuration. I mean, even for me, you know, if I get a system set up, I usually am like, Oh God, it's set up and I just leave it set up and I leave it I'm not doing what we would encourage a corporate user to do, which would be, apply, rinse, repeat, set up, destroy, until they've got everything perfect, and then that's their device. That mentality, to me, is rare and close from what you're describing, it would be you're going to treat it more like an appliance, unless you're an IT pro.

Claus Strommer

18:00

There is also, yeah, a different aspect that we haven't talked about yet on that is maintenance. I mean, as a home lover, and I kind of like, got thinking of this since what you mentioned, like, the like, try it until you get working home library is likely has a goal they on. Yes, they will tinker with the system until the restart goal, and maybe they might Tinker if they're dissatisfied with the result. But at a certain point, there's going to be a plateau where the effort of changing things is not going to be worth it anymore, so you stick with the devil let you know, essentially, an enterprise like it, that you really can't do that anymore. You have to have constant security patches and service updates and API changes. So it's a constant moving target, even if you wanted to stop unrest.

Rocky

19:14

Yes, and I think at least in the US, and I suspect, in number of European countries and even possibly China, a lot of folks are tech savvy, and they're worried about their privacy, and so they want to set up something where all their smart appliances and everything else doesn't report on them, and they have a network that they can feel secure in. And so yes, they want to get it. They would like an appliance, but the issue is keeping it maintained so they don't have security holes also. And I think there's a market for. That actually, because they're mostly tech folks that sit there and go, I don't want my TV listening in on me. I don't let because we know the dangers.

Claus Strommer

20:15

Yeah, yeah. And these are the kind of people who who start their home lab lab to like, run up high hole so that they can block certain ads on their TV or whatnot. And, yeah, the scenario of using like home assistant instead of Alexa or whatever. Yeah, I will.

Rob Hirschfeld

20:41

There are very few of those that are actually then set up, though, as you know, having real on premises or self managed infrastructure, because, because it's so hard, right? Go ahead, yeah,

Claus Strommer

20:55

I, I see a fair amount of demand for on premise, particularly on the assistant field. I think this largely comes from users having been burned by cloud based service providers like it, like if people running their ring cameras or whatever, and then they're being down because they're not their internet is down or because the cloud providers stopped supporting their hardware. People, yeah, and so on.

Rocky

21:38

They keep seeing ads related to searches they did on a different device, and they know that they got a bunch of don't follow me and other stuff in there, and they're going, what the hell? And they decide they're tired of it, and they they want more control or

Claus Strommer

21:58

straight service and certification. If, let's say, the vendor decides that a particular feature is not cost effective anymore and they drop it. Well, if you bought the device because of that feature, You're mighty disappointed, like, you know, the hardware can support it, yeah, so

Rocky

22:24

the let's break your equipment every three years, yeah,

Claus Strommer

22:27

or Roku, and on pretty much any smart TV, with their with their updates and So on, constantly getting worse, so that there's there I suspect that there's a large population of these home lovers who really like this is the same repatriation argument that we're seeing in the cloud, like and the cloud Is it's or in the enterprise, it's largely a matter of cost and sometimes sovereignty, but with home lovers, it's almost always like features that go missing. And while I wish the people who care about privacy, like there was more of them, I am also well aware that we are a niche of the general population, and while we are kind of in a privacy conscious echo chamber, I know very well that the large population, even the home lovers, for them prices, is not really that the most important factor is really just convenience and features.

Rob Hirschfeld

23:53

Yeah, what's interesting is the way you're describing it, thinking back to the original question, is you're still describing an independent market. You're talking about it people who want something for their home, who are right, but that is, that is a market where they are doing the work. They're not looking for an enterprise product that they're using out of, you know, off book, if you will, or as a community product. They're, they're looking for a home experience.

Rocky

24:31

Well, yes, I think you have to expand beyond the IT professionals, science professionals, the science folks are really heavy into software and whatnot, but they expect their software to be tools, not projects, and they have a lot of the same concerns with what's going on right now in uh. Are protecting their community, if you will. So it's expanded beyond the IT professionals, but who, it's still a

Rob Hirschfeld

25:11

market, I guess, is sort of like, so, so the right, this is part of the reason for this topic is we have a dilemma all the time of like, does RackN spend time encouraging home use? And so, you know, we are, we are building an enterprise product. So what you're describing, what this conversation leads me back to, the No, the people who would use us at home are and would be like, Oh, this is great. I want to automate all this stuff at home. We're not going to fit their use cases at all, or there were the usability for them. The only time somebody would do it is if they're just it's a learning model for them. And so, and I, but I'm, I think I'm generalizing after this conversation, which is fair to say, yeah, if you're doing something from a enterprise, it perspective to solve enterprise problems, even if it could work at home, it's probably not designed for that to be usable in that case, yes,

Rocky

26:21

yes, and there needs to be a home solution too. So it's Yeah, and our enterprise is not particularly fit for use, but at some point somebody will come up with a home solution that's similar to the passport password managers and the ad blockers and everything else like that. And when that comes out, a lot, lot of the the home computer folks, not necessarily it. Folks will jump on the bandwagon because they would like to be able to manage their router and their internal net and keep it separate from the cloud and stuff like that. But it's too much effort at the moment.

Rob Hirschfeld

27:10

Yeah, no, it's it's really hard, and ultimately, it's probably going to need to be a more integrated system than than the current offerings. Yeah, even

Claus Strommer

27:21

if it was integrated, I don't know if there would be a demand for home use, like, how many? How many people do you see using MDM to manage their family cell phones? No one, practically, no.

Rob Hirschfeld

27:35

It's the overhead is a huge, is a huge burden on that

Claus Strommer

27:42

you might end up just buying into a vendor ecosystem like Googles or apples, because it lets you manage enough of your devices across central location. For example, with Google I can, I can manage like, it's like screen time and I don't need anything else. Are there third party solutions for this? Yes. Do I trust them? Mostly, no. Trust Google? No, I wish. I wish I didn't have to, but,

Rocky

28:25

but you don't trust them, but you still use them, but you know you can't trust them.

Claus Strommer

28:29

Yeah, yeah. So the moment I stopped needing to do to manage my kids devices, because they're old enough, I like that, that vendor lock in goes away.

Rocky

28:45

Yes, I can the kids problem is a as a big one, and that that's one where and, and some ways this harkens back to the old turn key firewall systems. So yeah, people want them there. The the techies want them from home for home, but it's still there are lots of small businesses that actually would pay for a turnkey system that manages their small computer network or their medium computer network, so smaller than enterprise, but larger people with more money than the home brew computer folks,

Rob Hirschfeld

29:34

I can tell you that It's incredibly hard to make any it doesn't make financial sense, let me say it that way. To support a retail to support a real retail transaction on a something that has that much variability is the ultimate problem, right? It works from an open source like PF sense. Yeah, because at the end of the day, the open source person doesn't have any obligation to make sure it works on whatever you're installing it on. You hope it does, and community sort of will polish that stone. But you know, if you're dealing with a business and they're, you know, buying cheap stuff and then trying to make it work. It's the economics of supporting that are incredibly hard. I mean, this is, this is our challenge, right? If we do home loving stuff, and you're trying to boot pies or nooks or something like that, and give us, you know, give us a call to get help with that, you know, there's, there's no, you know, there's no, there's never going to be an economic return on helping somebody solve that problem. Now, I'd love to have a community of people who are excited and bootstrapping. That would be cool. But you know, at some point it's going to cause, you know it, you have to invest in in doing that. And then, if there's no crossover, then, right? If it, you're either doing it for Goodwill, or there's, you know, some, some other supporting

Rocky

31:15

capability, yeah, no, your company it, it doesn't make sense for you with your overhead and your structure and whatnot. What it makes sense for is consultants and stuff that who have come up with a solution, that if a small business wants something, they'll go in and consult and do what, what they need for that company,

Rob Hirschfeld

31:40

if it's, if it's an overall solution, and they're, they're having a repeatable piece, and then they can go into that person, say, you need to buy these servers, this is what we support. And then they're providing other services. And when they show up on site, they're getting paid for it. All of those things stacked together to create a commercial, a viable commercial solution, right? Totally, totally agreed. And but this is, it's always a struggle for us, because, you know, we want people to be, you know, the, you know, these, these home laborers are enthusiasts, and they're doing cool stuff, and they're usually us. And so we're like, yeah, we want to make that happen, and we want to make it work, and then it can be really frustrating when you know, somebody's, you know, trying to do that, or, you know, and we're we just don't have we can't, we can't help them the way we want

Claus Strommer

32:36

for us. Yeah, go ahead. Yeah. It would really be nice to be able to see a solution for RackN or the for the home use? Yeah, I think that given time it would do, would be a good demand in the home lover community. But of course, like in the time it could, you could do like get traction and for for people to actually become familiar with it, and that time you, you're, you're not making a profit right From it, or even in the long term?

Unknown Speaker

33:22

No, it's

Rocky

33:24

volunteer, developer, advocate for that niche.

Rob Hirschfeld

33:31

No, and there's, that's, you know, you would, what I always hope is that there's enough. We're doing, enough to help. You know? We just pull along a community component, right? Because we're doing all the other work. But, you know, just like Kubernetes, is complex for reasons, right? We're, we have very little incentive to streamline something for that use that this, right? We're back to the challenge, right? It's like, you know, if our if, if you want to do the Mack truck as your daily driver, it will really haul a lot of cargo for you, but it's probably overkill for that, that task. And so

Unknown Speaker

34:14

everybody wants to blame Broadcom.

Rob Hirschfeld

34:18

There was the same, there was a, yeah, there's a whole bunch of Broadcom like, I have seen home labbers who are buying, like, real gear and setting up the V center stuff and things like that. But once again, they weren't doing that for the use cases, the home use cases. They were doing that as professional, by and large as professional training. Even the Kubernetes people seem to be doing it out of professional need, like even the PI clusters that I saw and things like that, they were mostly people playing with Kubernetes, not solving a home, you know, a problem for themselves. So. Only there would have been easier ways to solve a problem, the cluster of pi Cooper of Kubernetes by Raspberry Pi's, much as Oracle wanted to show you 1000s of them running it

Unknown Speaker

35:14

a silly use case. All right. All right.

Rob Hirschfeld

35:17

Thank you all for indulging me in this and this, it was interesting to hear everybody's point of view. Maybe one day we'll find a find a home lab use case, right? Home product is what we really ended up towards. Cool. Next week, I'm going to try and wrangle the team and talk about some of the adventures in Kubernetes virtualization that we've been having. So just have to give them some warnings so they're

Speaker 2

35:48

not double booked. Sounds fun. Definitely. Okay. Thanks. Thank you, Rob.

Rob Hirschfeld

36:01

This is one of those fun asides where we sort of dive into a topic and scratch an itch that we all see and have we do edit these out. And there was a whole bunch of preliminary conversation talking about KubeCon, which is going on at the same time, security vulnerabilities out of work, cloud and other topics of the day, those are fun and interesting conversations too, and you're missing out because you're not in the live conversation. So I would encourage you to join us in one of these Tuesday afternoon or potentially on a Thursday strategy session. You can find more details out at the 2030 dot cloud, and I'll see you there. Thank you for listening to the cloud 2030 podcast. It is sponsored by RackN, where we are really working to build a community of people who are using and thinking about infrastructure differently, because that's what RackN does. We write software that helps put operators back in control of distributed infrastructure, really thinking about how things should be run, and building software that makes that possible. If this is interesting to you, please try out the software. We would love to get your opinion and hear how you think this could transform infrastructure more broadly, or just keep enjoying the podcast and coming to the discussions and laying out your thoughts and how you see the future unfolding. It's all part of building a better infrastructure operations community. Thank you.

00:0000:00