Artwork for podcast Data Driven
Bulletproof KDE and Why Linux is Winning Over Developers
Episode 415th June 2026 • Data Driven • Data Driven
00:00:00 00:58:15

Share Episode

Shownotes

On this episode of Data Driven, hosts Frank La Vigne and Candace Gillhoolley are joined by hardware and open source expert Michael Makowski to discuss the shifting landscape of developer workstations and AI hardware. As Windows usage declines among developers and AI engineers, Linux is experiencing a surge in desktop adoption. Michael takes us inside the latest efforts to make Linux not just accessible, but enterprise-grade—sharing how his team is driving advancements in stability, reliability, and user experience for validated Linux hardware.

We talk about the dramatic improvements in Linux desktop support, the importance of privacy and avoiding surveillance-driven proprietary systems, and the game-changing features coming to market—like automated system rollback and curated app installs. Plus, we explore the current state of gaming on Linux, the technical edge unified memory brings to AI development, and why companies are increasingly opting for supported, Linux-based workstations. Whether you’re Linux-curious, rethinking your hardware choices, or just passionate about the future of developer tools and data engineering, this conversation will equip you for what’s next.

Links

Time Stamps

00:00 Website security concerns and solutions

05:03 Supporting KDE for long-term stability

09:45 Desktop environment compatibility issues

10:46 Conflicts in desktop environments

15:07 AMD vs Intel & Nvidia Performance

18:58 Showing the production site

23:49 Steam's Linux runtime environment

25:43 Running Windows games on Linux

29:47 Concerns about software privacy issues

33:31 Migrating from Windows challenges

37:48 Setting up machine learning hardware

41:31 Resolving system issues efficiently

42:59 Setting up a VPN correctly

47:40 Running VMs on alternative OS

52:49 Upcoming OS Upgrade Details

56:19 Rigorous testing and development process

57:25 Tuning BTRFS for performance

Transcripts

Speaker:

Today it's 5%. It's two and a half times greater.

Speaker:

And I think that we're reaching a tipping point. So

Speaker:

to meet that tipping point, we also need to match the

Speaker:

expectations of people who are coming from commercial software and

Speaker:

quite frankly, up the game and up the support

Speaker:

mechanisms and provide the capabilities the proprietary users

Speaker:

are really looking for. And that is stability and reliability

Speaker:

and validation. Hey, Frank here. I just want to give a little bit of

Speaker:

context for today's episode. It's basically not really a

Speaker:

data or AI talk per se, but it is something

Speaker:

that I think that the community in general needs to start thinking

Speaker:

about. The Windows that we know and love

Speaker:

isn't really a thing anymore. And increasingly I'm seeing

Speaker:

a lot of workloads and developers and AI

Speaker:

engineers shift away from Windows into

Speaker:

the Mac or Linux ecosystems.

Speaker:

Now, what this means for you practically is. I get it, Linux can

Speaker:

be a little intimidating at first, particularly on the desktop, but

Speaker:

we have the opportunity to speak with someone who is

Speaker:

really alleviating that. So listen with an open mind and keep

Speaker:

in mind that the future may not have a start menu on it.

Speaker:

Hello and welcome back to Data Driven, the podcast. We explore the emerging field of

Speaker:

data, AI and data engineering. Now,

Speaker:

Andy couldn't make it today, so with me, I have the most

Speaker:

quantum curious person I know. But today we'll be talking about AI and data

Speaker:

and hardware. So who do we have talking to us today? Candace?

Speaker:

Today we're going to be speaking with Michael Makowski, and

Speaker:

he is a really fascinating guest because he sits at the

Speaker:

intersection of user experience, open

Speaker:

source development, Linux and the evolution of web

Speaker:

applications. He is an author of single web

Speaker:

applications and a longtime advocate for SPA

Speaker:

architecture. And I'm really, really excited to dig in

Speaker:

deep about what he's doing now. Very cool. Well, welcome to the

Speaker:

show, Mike. Hey, thanks. I appreciate it. No problem.

Speaker:

Spa. So you like to spend time at the spa? Well,

Speaker:

I did back in the day, although recently

Speaker:

with the introduction of AI exploitation

Speaker:

tools. They're very popular in Hollywood, from what I understand,

Speaker:

although maybe they're applied a different way there. But as far

Speaker:

as websites are concerned, the more active your website

Speaker:

is today, the more likely it is to get

Speaker:

exploited, especially if you're using something like WordPress. I

Speaker:

spent some time in the security space,

Speaker:

actually about four years in Silicon Valley, and

Speaker:

the number one attack vector for a

Speaker:

lot of companies was WordPress.

Speaker:

So our website actually is all static these

Speaker:

days. So after writing an spa book, and there are great

Speaker:

things you can do with an spa, basically Having a web application as

Speaker:

opposed to a website. But for

Speaker:

what we're doing today, which is selling validated

Speaker:

Linux systems with a fully fleshed out

Speaker:

web infrastructure, we don't have a need

Speaker:

for having a web application. So

Speaker:

you don't have a need for it and you value

Speaker:

privacy and security then. I'm a big fan of

Speaker:

doing static web pages, which is what we're doing now.

Speaker:

Interesting. So you, you make and ship

Speaker:

Linux laptops. Is that high performance Linux laptops? Is that a

Speaker:

correct statement? Yes, we actually do arrange.

Speaker:

We're launching this week. Hopefully we'll get everything

Speaker:

pulled together. Major initiative which we've

Speaker:

actually had under the way. We're now going into our second month on that

Speaker:

and then we're launching, at least depending on

Speaker:

how you want to count it, six new or refresh

Speaker:

models. So with the latest

Speaker:

2.26.04

Speaker:

Kubuntu operating system, which we

Speaker:

provided stipends for the developers and then directly help

Speaker:

develop and also we provide hardware and then

Speaker:

we're a patron for kde, so we get to pay a lot of

Speaker:

different ways to help make sure that we've got a really

Speaker:

reliable system. And then the

Speaker:

other way that we are moving the needle forward

Speaker:

for validated Linux systems is by

Speaker:

announcing a bulletproof KDE

Speaker:

initiative. That's our working title. We don't know exactly if that's

Speaker:

going to be what sticks for the actual press releases, but

Speaker:

that's a $100,000 plus commitment

Speaker:

to KDE for the next three years to ensure that we

Speaker:

have a stable long term support release on

Speaker:

KDE the desktop. And this is going to

Speaker:

benefit lots of companies and organizations that

Speaker:

use kde. A great example. You'll see that if

Speaker:

you look around NASA that the desktop they offer now running is KDE.

Speaker:

And of course we have Fortune 500 companies

Speaker:

that use our systems and AI companies

Speaker:

and all those folks are going to benefit. But basically anybody

Speaker:

that uses KDE and wants to use the LTS

Speaker:

release will get a big boost. And then of course

Speaker:

we do have a little, what is it, enlightened self

Speaker:

interest involved with this because you know, we're funding that. So hopefully it's going to

Speaker:

drive interest to the systems we produce

Speaker:

and so that we can build a virtuous cycle so that

Speaker:

we can continue to build an ecosystem that

Speaker:

meets or exceeds the best proprietary systems

Speaker:

available, but without all the spyware and without

Speaker:

all the surveillance and all the phoning home and

Speaker:

the age verification APIs that are basically

Speaker:

a thinly veiled attempt to have a

Speaker:

national ID system in place. You know, a lot of people

Speaker:

are really recoiling against all the Big Brother activities

Speaker:

that are occurring currently. And we

Speaker:

have recoiled against that for a long time. But

Speaker:

it's getting so overbearing these days that we're seeing

Speaker:

an extreme amount of interest. Linux usage just a few years

Speaker:

ago was 2% on the desktop, today it's 5%,

Speaker:

it's two and a half times greater. And I think that we're reaching

Speaker:

a tipping point. So to meet

Speaker:

that tipping point, we also need to match the expectations of people

Speaker:

who are coming from commercial software and quite frankly up the

Speaker:

game and up the support mechanisms and provide the

Speaker:

capabilities the proprietary users are really looking for.

Speaker:

And that is stability and reliability and validation.

Speaker:

And so Candice has been involved with me on this for a

Speaker:

seven year journey and she knows that we've been beating that drum

Speaker:

for a long time. But I think that sometimes you just keep on doing

Speaker:

what you're doing and then, you know, the whole world shifts

Speaker:

and you're there to take advantage of it. So I get a

Speaker:

feeling that we're around that zone right now. No, 100%.

Speaker:

And I think that, you know, the privacy story as you look at, you know,

Speaker:

and most of our listeners are Windows users, I would say

Speaker:

with some Linux experience. So KDE is the desktop environment,

Speaker:

is that. Yes, that runs on top of any distro

Speaker:

Linux or just Ubuntu or just certain ones.

Speaker:

It runs on virtually any build

Speaker:

of Linux. Okay. But the.

Speaker:

It's one of those things that's in a certain extent like

Speaker:

Windows in that it doesn't like to

Speaker:

share the OS with other desktops.

Speaker:

Right. Actually, that's kind of the wrong term. It's more like the other way around.

Speaker:

Certain desktops don't play well with other desktops. So like certain

Speaker:

kernels don't play along with certain desktop

Speaker:

experiences. No, kernel is very lower

Speaker:

level. Yeah, well, not lower level. It's. But it

Speaker:

is very

Speaker:

generic in the respect that it's a component

Speaker:

that everybody works with. What you run into trouble with

Speaker:

is, you'll remember back in the day when they had

Speaker:

ZR dos, you know, I'm going way back here, so

Speaker:

that's a lot goes way back. Yeah, yeah. We're losing about three quarters of your

Speaker:

audience right now. But bear with me, back in the day,

Speaker:

Gary Kildall from Digital Research created

Speaker:

ZR dos, which was dos but with stability

Speaker:

and performance. And it was better than

Speaker:

Ms. DOS by almost any

Speaker:

measurable factor. And it gained a lot of

Speaker:

popularity. The problem was that Microsoft

Speaker:

didn't like it. So what Billy and his friends did

Speaker:

was they decided to make it incompatible

Speaker:

for their Windows program. Windows 95

Speaker:

ran perfect on it until they decided to make it not run perfect.

Speaker:

And so it's. Now, not everybody

Speaker:

is so nefarious as that. But if

Speaker:

you don't build your desktop system to work with

Speaker:

the underlying layers, you're going to have problems or if

Speaker:

you create like services that conflict. And

Speaker:

that's what indeed can happen with

Speaker:

certain other desktop systems in Linux.

Speaker:

So the most notorious one, and I'm not saying that this is

Speaker:

an example of any sort of

Speaker:

malfeasance, it's just that the GNOME

Speaker:

desktop environment has a lot of services

Speaker:

that are required by GNOME that

Speaker:

don't play well with other desktop systems.

Speaker:

And I'm certain it goes with other ways as

Speaker:

well. But. So that's the reason why you see distributions that, that only

Speaker:

ship one or two desktop environments. Instead of

Speaker:

giving you 20 to choose from if you want to like go out and play

Speaker:

with a bunch of them, it's because they conflict. And

Speaker:

again not because people have malfeasance in their heart, but

Speaker:

because of the fact that nobody actually tests to make

Speaker:

sure that the interplay actually is going to

Speaker:

work out the way you want it to. I think we've probably done

Speaker:

a lot more than the average bear in that respect where we've actually identified

Speaker:

the conflicting services, but nobody's paying us,

Speaker:

quite frankly, to make those

Speaker:

conflicts go away. Or nobody's paying Gnome or

Speaker:

KDE to make sure those conflicts go away. There's really not

Speaker:

much of an incentive to change that because people who

Speaker:

use a certain desktop tend to stick with it. It's kind of like they

Speaker:

choose a type of cigarette and they stick with it too. Maybe not, maybe not.

Speaker:

The best example, they picked a favorite type of soda. How about

Speaker:

that? Yeah, there we go. Exactly. Pepsi or Coke? There we go. Pepsi or

Speaker:

Coke or Dr. Pepper. Yeah, so

Speaker:

what? I think you're absolutely right. I think there's a.

Speaker:

Something is rotten in the state of Windows land. We have a lot of folks

Speaker:

here who are Microsoft centric, who do listen. And

Speaker:

if they've been following me, they know that I've kind of gone. I've gone

Speaker:

rogue, right? Because Windows 11

Speaker:

has really pushed me over the edge in terms of the decisions that were made,

Speaker:

the usability and when you start peeling back the

Speaker:

curtain behind the privacy story on Windows 11, certainly the home version

Speaker:

Pro is not as bad. Enterprise is a little bit less bad. It's

Speaker:

painting a picture. I really don't like. Right. My, my beef was, you

Speaker:

know, if you look at my office, I got a lot of equipment and

Speaker:

you know, because of an arbitrary decision made in Redmond,

Speaker:

most of this will not run Windows 11. And

Speaker:

that really made me mad. I was going to use other words, but

Speaker:

I think Windows 11. And again, I used to be a Microsoft

Speaker:

MVP. I used to be a Microsoft employee twice

Speaker:

and I doubt that I'll be reinvited for the MVP program

Speaker:

anytime soon. But I just take issues with

Speaker:

some of the decisions that were made with Windows. Now I haven't seen what

Speaker:

they announced at build this week but or last week,

Speaker:

but I'm just very much annoyed and

Speaker:

I think that Windows choices,

Speaker:

be they what they are, I think has really driven Linux

Speaker:

adoption. You mentioned Linux on the desktop has

Speaker:

doubled. I don't think it's a coincidence that Windows 11 has drove

Speaker:

away a lot of hardcore Windows users like me. Right. Like, I mean I,

Speaker:

for me to, you know, obviously I, you know, my day job is at Red

Speaker:

Hat now. You can see the fedora there. Right. But you know, for the longest

Speaker:

time I was the only person on my team that had a Windows laptop.

Speaker:

Right. But with the rise of AI development tools

Speaker:

being fairly agnostic in terms of what

Speaker:

platform they're on, almost to the point now where they're, they recommend

Speaker:

Mac or Linux and if you have a Windows device they

Speaker:

actually say you probably should run this inside WSL or Windows subsystem for

Speaker:

Linux. I think that you're right. I think we are on the cusp

Speaker:

of an inflection point of Windows.

Speaker:

Certainly in the AI field, if you're an AI developer, it's almost

Speaker:

exclusively Macintosh, I would say.

Speaker:

Well, we have a lot of companies

Speaker:

that develop for AI that use our systems.

Speaker:

Right. There's a saying in Silicon Valley

Speaker:

which is nobody deploys to Mac.

Speaker:

So when you really think about it, the

Speaker:

tools have to first and foremost be

Speaker:

Linux compatible and Linux proven and

Speaker:

Linux battle tested. Right. Mac is popular because

Speaker:

got a trophy status and they also are

Speaker:

popular now because of the unified memory mechanism. They

Speaker:

have relatively weak GPUs compared to the

Speaker:

top end Nvidia GPUs obviously. But because

Speaker:

they have unified memory that can overcome that

Speaker:

disadvantage and for certain models can actually run faster.

Speaker:

Now with the advent of AMD processors as

Speaker:

well as now the latest Panther Lake intel

Speaker:

processors, Those now have BPIGPUs in them as

Speaker:

well. Again, they don't touch the Nvidia in

Speaker:

terms of the level of performance

Speaker:

available. But again for certain models, say

Speaker:

running llama from my understanding, and I'm not an expert here, so if

Speaker:

I sound foolish, it's because I am.

Speaker:

But the. For certain

Speaker:

models, Llama, for example, will work better because you can

Speaker:

have a memory window of 128

Speaker:

gigabytes, for example, instead of being limited to 16 or 24

Speaker:

gigabytes on an Nvidia system. On the other hand, you know when you got a

Speaker:

5090 and you want to start swapping stuff out, if your model is

Speaker:

the right structure, that 59 is going to decimate anything

Speaker:

else. But I think the reason why you're seeing people developing

Speaker:

on Mac for models that will eventually be

Speaker:

deployed, typically on Linux Nvidia

Speaker:

systems, is because of the fact that they,

Speaker:

they can have a

Speaker:

proprietary quality,

Speaker:

user experience and reliability and

Speaker:

then also have reasonable performance when working with their

Speaker:

model. Sometimes better performance working with their model,

Speaker:

although again, I don't think that tends to be deployed. But I'm

Speaker:

not an expert in the deployment arena. I just know what our customers

Speaker:

buy and what they tell me. Well, no, I mean,

Speaker:

so as someone who is in AI space, I will say that

Speaker:

unified memory, I think is what makes the Apple Silicon really appealing.

Speaker:

And I also have a DGX Spark, which that's also. It's kind of claim to

Speaker:

fame is that it's unified memory, right? Oh yeah. And you're right, like depending on

Speaker:

the actual chip you can get more done,

Speaker:

but very often the VRAM that's

Speaker:

available to you or the RAM that's available to the video card

Speaker:

is also a blocker too. Right. So the fact that you can have a unified

Speaker:

memory space I think is a huge deal. And you're even seeing Microsoft follow

Speaker:

suit with this with their new I think RTX Spark. I think that's what it's

Speaker:

called. Yeah, I think that's kind of the future.

Speaker:

But yeah. So tell me about some of the AI specific hardware.

Speaker:

You have me curious. Well, I can show you what we have that

Speaker:

people. Yeah. And then we'll let Candace ask, ask questions because I'm hogging away. We

Speaker:

have lots of companies that buy hardware

Speaker:

that. And I'll show you the heavyweight stuff.

Speaker:

Let's see. Yeah, we're actually, like I said,

Speaker:

going to be launching new versions of these systems. But

Speaker:

I'll, I'll see if I can share a.

Speaker:

Let's go to our website, be able to share your screen and I'll,

Speaker:

I'll talk through it for those that are listening.

Speaker:

Yeah, you just go, ooh, wow, you can do

Speaker:

that. Take my money. Yeah, that's my recommendation.

Speaker:

Yeah, take my money. Yeah, you can keep doing that.

Speaker:

Let's see. Share and let's try

Speaker:

a home tab. There we go. And there we go.

Speaker:

If those who are listening and they want to see this, we are on YouTube.

Speaker:

Just check the show notes. I always link the YouTube version. Sorry, cut you

Speaker:

off. That's it. No, no, you didn't cut me off. Interesting thing about doing

Speaker:

these web pages, by the way, is that I

Speaker:

think it's a pretty fascinating thing here. But if you look at these

Speaker:

graphics here, right? And then I'm going to change this to a

Speaker:

dark theme right now. So the OS is changing to the dark theme

Speaker:

and. Oh, that's on a local host. Let's do it. Let's go to the actual.

Speaker:

I was showing the development model. That's the reason why that broke.

Speaker:

So here's the actual production site. But what's pretty

Speaker:

fascinating about this that you never think about is that these are

Speaker:

mostly black, you know, very dark machines, right. But

Speaker:

if you now switch over to a light view, you

Speaker:

may not notice it. But these, the chassis here just bumped up

Speaker:

about two levels of lightness in the gray area.

Speaker:

And it's a kind of funny thing because the, you

Speaker:

put something in a white room and it makes the whole thing lighter.

Speaker:

Those things you don't just run into unless you're, I

Speaker:

guess, selling black computers on the Internet. So let's take a

Speaker:

look at a couple of the products that we have and the ones that are

Speaker:

popular for AI. So this is the, this is

Speaker:

the beast. And it is a 5090

Speaker:

with 24 gigabytes of video RAM. Oh, wow.

Speaker:

Yeah. And then it's got up to 192 gigabytes of

Speaker:

RAM. So I guess 2000, $2500 of RAM

Speaker:

at wholesale these days that you put into that. It's got an 18 inch

Speaker:

screen, 2560 by 1600, which is a beautiful

Speaker:

screen. It's DCIP3, 100% and

Speaker:

it's just a beast. It's got a subwoofer in it for, you know. Oh, wow.

Speaker:

Crying out loud. There we go. I almost slipped up there.

Speaker:

So. And the, the subwoofer does sound good. So

Speaker:

you can get your jam on when you're loading up

Speaker:

your, oh, I don't know what 30 billion

Speaker:

parameter model. See, I might be sounding ignorant here now too,

Speaker:

but yeah, while you're loading up your model and this one's real popular

Speaker:

with ML. What's the,

Speaker:

what's the 5090 vram?

Speaker:

24 gigabytes. Oh yeah, you can do, you can do some serious damage with

Speaker:

that. Yeah. So.

Speaker:

And it's got the 24 core. The new version coming

Speaker:

out is the. And this was actually recommended

Speaker:

by Wired magazine as the best Linux

Speaker:

laptop for performance. And the next generation is

Speaker:

the 2090 HX which gets it about a 6 to

Speaker:

15% bump in single and multi core

Speaker:

performance. The GPU remains the same,

Speaker:

but that gives incredible performance as well. So pardon

Speaker:

my ignorance, but one of the, one of the screenshots that kind of cycled

Speaker:

through was a game. Yes, obviously,

Speaker:

obviously anyone buying this would buy it for serious business purposes. But what

Speaker:

is the state of gaming on Linux these days? I've

Speaker:

heard things kind of mixed and is Kubuntu also what is

Speaker:

based on the Stream Deck or Steam Deck is

Speaker:

not what the Steam deck is based on. Steam Deck is

Speaker:

a custom version of Arch

Speaker:

Linux that uses a non

Speaker:

mutable version. It's designed to be

Speaker:

an appliance, not a general purpose computer.

Speaker:

And there are some alluring aspects of that.

Speaker:

Obviously if it's designed to be an appliance, that means that the end user

Speaker:

is unable to easily break it. The

Speaker:

downside to that is that the end user is

Speaker:

unable to easily add system level services

Speaker:

to the system. So. So if you want to run a database for example,

Speaker:

and have a multi user system, well, you're probably kind of out of luck. But

Speaker:

then again you're running a Steam Deck. You're not. It's something that you hold in

Speaker:

your hands. Well, it's like an Xbox, right? Like a PlayStation.

Speaker:

Yeah. They by the way run KDE Linux and

Speaker:

I'm sorry, KDE Linux, they're running the KDE desktop on that as well.

Speaker:

Gotcha. Okay. That's where I saw it because in case people are wondering what's Frank

Speaker:

talking about? And again I'm not a Linux guru. I, one of my

Speaker:

co workers said I speak Linux with a Windows accent. The. When you

Speaker:

go to kde.org, one of the devices on there is. You're, you're on that

Speaker:

page and. But next to your logo,

Speaker:

a couple of items down is, is the Steam deck. So I

Speaker:

noticed that like. And I noticed that Linux gaming has really taken

Speaker:

off. Right. It used to be like good luck with that.

Speaker:

I think it used to be. Now it's. Now I would say again this is

Speaker:

just based on conjecture and I may sound like an idiot now, but I would

Speaker:

say if you want to play a triple A game, it's probably a

Speaker:

5050 shot that it will run on Linux. I

Speaker:

think it's better than that even really. At least if it's a. If

Speaker:

it's a triple. If it's a triple a game

Speaker:

like if you want to run something like Grand

Speaker:

Theft Auto, one of my favorite Grand Theft Auto 5.

Speaker:

So almost certainly you can run that.

Speaker:

Okay. The biggest challenge right now with

Speaker:

gaming on Linux is not a technical issue

Speaker:

because because

Speaker:

of Steam's investment

Speaker:

in that aspect, they've created a

Speaker:

runtime environment that is not a

Speaker:

translator, but is an

Speaker:

API layer that Windows programs can run

Speaker:

and say, I want to make this video call. And it

Speaker:

works the same on the Linux system as it does on

Speaker:

Windows. So it's not running through a

Speaker:

like Rosetta Stone translation layer. It's

Speaker:

running in a virtual environment that then makes

Speaker:

calls out to local

Speaker:

resources and they just work. So

Speaker:

if you can imagine you in Windows, and

Speaker:

this is a very simple example of this, and if any

Speaker:

of your readers want to pick apart my explanation, have at it. But

Speaker:

I think in a grand sense this is correct.

Speaker:

When you're on Windows and you want

Speaker:

to list the files in a directory, you type in dir. Correct.

Speaker:

And when you're on Linux and you want to type or list files in a

Speaker:

directory, you use ls. And the layer

Speaker:

which is called WINE that is used by

Speaker:

Steam, it basically aliases

Speaker:

LS for dir. And so. So that's what WINE is.

Speaker:

Okay. In fact, WINE is something along like

Speaker:

Wine is not an emulator. It's one of those recursive

Speaker:

acronyms, I believe, that plague us to this day.

Speaker:

Find is not elm. Yeah, yeah, stuff like that.

Speaker:

The whole point is that it's a layer that instead of

Speaker:

creating a whole new emulation environment and having to recompile

Speaker:

stuff, it just takes the executable and allows it to run.

Speaker:

And then when it asks for a DirectX call, there

Speaker:

is a. Actually a repointer that occurs

Speaker:

that allows that to work. I think DirectX

Speaker:

has fallen out of favor in for a lot of

Speaker:

games they just go right to Vulcan. And Vulcan is native to

Speaker:

Linux, so there doesn't even have to be a translation. There are

Speaker:

a number of AAA games that actually run faster on Linux than they do on

Speaker:

Windows. Especially now that you have all those background services running

Speaker:

on Windows 11 to take snapshots and send them to Lord knows

Speaker:

where or to drive, you know, you have to have that background service so that

Speaker:

you can get advertisements in your system settings

Speaker:

configuration or get advertisements in your Start menu

Speaker:

and you know those are important. So you can't turn Those off. No,

Speaker:

yeah, yeah, people. And that's actually driving

Speaker:

Linux to be more power efficient. We're able to get longer

Speaker:

battery lives on Linux, then you are able to get on

Speaker:

Windows, even though Windows has such

Speaker:

fantastic cross device driver support.

Speaker:

You can't bring the CPU necessarily

Speaker:

down to the sleep level that you want it to if it's

Speaker:

constantly calling all these background services. So

Speaker:

through some tweaks we're able to be competitive or even better than

Speaker:

Windows as far as battery usage and battery life on

Speaker:

a number of our systems. So you've already

Speaker:

listed a couple of the truly competitive advantages

Speaker:

for someone who's considering gaming the gaming

Speaker:

platform they want to use. That's exciting. It's

Speaker:

actually very exciting. What else do you think would surprise people the most

Speaker:

about the current state of game compatibility, performance and

Speaker:

overall user experience when considering

Speaker:

Linux as their primary gaming platform? Well, I do want

Speaker:

to answer one thing that is a negative

Speaker:

for Linux and it's not because of Linux, it's because of the game companies

Speaker:

and the Sweeneys of the world. And that is

Speaker:

what's a Sweeney? Sweeney runs

Speaker:

the company, I believe, that creates Fortnite.

Speaker:

Oh, okay, okay. So you've got on one hand

Speaker:

the Steam team that

Speaker:

wants to get out from under the yoke of Microsoft and they

Speaker:

went and did something amazingly

Speaker:

proactive with releasing their Steam deck running Linux,

Speaker:

not Windows, you have

Speaker:

that side of it and then the other side you have game companies that want

Speaker:

to make sure that their intellectual property

Speaker:

locked down and the weather locking that down is by using

Speaker:

kernel level anti cheat mechanisms.

Speaker:

And from my understanding what that means is that

Speaker:

they're actually running in kernel space

Speaker:

code that is meant to prevent people from

Speaker:

cheating on massively multiplayer

Speaker:

games. So mm pgs,

Speaker:

you can tell I'm a real gamer, but that is a downside that

Speaker:

MMPGs are not always as well

Speaker:

supported or not supported officially by the game studios.

Speaker:

And the on the flip side,

Speaker:

the fact that you're running code at the kernel level on

Speaker:

your Windows system is going to be an

Speaker:

incredible vector for somebody probably real

Speaker:

soon to own a lot

Speaker:

of Windows systems. Right? Because

Speaker:

people who just want a game are going to download whatever, they're not

Speaker:

going to go and review the source code

Speaker:

for the kernel code that's running in Windows. And

Speaker:

Lord knows you probably couldn't do that if you wanted to because it's certainly

Speaker:

protected under 3 NDAs and a couple of

Speaker:

licenses that have been created for that

Speaker:

purpose. And so you're just

Speaker:

basically adding mystery code on top of a system

Speaker:

that is built for surveillance. And so now

Speaker:

what's going to happen when you get into that and then what's going to happen

Speaker:

when commercial OSes are going to be

Speaker:

required to effectively tap into a national

Speaker:

ID database for the save the Children

Speaker:

effort for age verification? That's their vector.

Speaker:

Now you're going to have somebody running so they can play their

Speaker:

favorite mmpg. They're going to have running kernel level

Speaker:

code that somebody with Claude code is going

Speaker:

to be able to dissect and find a backdoor for.

Speaker:

Once they get into the kernel level they own the system. Now

Speaker:

you have all these wonderful tools and APIs available at

Speaker:

the the kernel level and the user space level that allow

Speaker:

you to go in and basically docs anybody

Speaker:

And I don't want to be part of that. So I guess the

Speaker:

short or the long answer because I've made this long

Speaker:

is that you can't game everything because

Speaker:

of these kernel level anti cheat as well as other anti

Speaker:

cheat mechanisms that you can't get anything but Windows.

Speaker:

But you have to ask yourself do you really want to be

Speaker:

putting those things on your system because of the security

Speaker:

implementation. Sorry,

Speaker:

implications, Implications. There we go. My

Speaker:

mind was trying to throw in an extra syllable there but the security

Speaker:

implications that you're exposing

Speaker:

yourself to so sounds dangerous.

Speaker:

Yeah, you're right.

Speaker:

I think the whole national ID thing is a

Speaker:

whole other can of fish can of worms to open.

Speaker:

But you're right just from an operational security point of view, if these

Speaker:

things are running in kernel mode. For those who don't know kernel mode an error

Speaker:

in kernel node is what caused the crowdstrike issue. That was

Speaker:

two summers ago and there are very good reasons

Speaker:

to run things in kernel mode. But I have a hard time thinking that the

Speaker:

risk of

Speaker:

cheat prevention. I don't know if it raises to that level in my

Speaker:

opinion. Right. And I think.

Speaker:

Do gamers. Are gamers aware of this or. And I know I don't want to

Speaker:

make this all about gamers because even though it's cool but these are high performance

Speaker:

systems. Right. But so sorry about that but do gamers

Speaker:

know the risks involved? I

Speaker:

don't think all of them do. Certainly some do.

Speaker:

Right. I would say that people that have

Speaker:

been ardent Windows users

Speaker:

for a long time like Linus Text tips. Right, right. Right

Speaker:

out of Canada they just did their Linux challenge which seems to

Speaker:

be like a yearly thing for them now or once every few years anyway.

Speaker:

I've done a few of them now and maybe

Speaker:

three. I don't follow that closely but

Speaker:

this last Time around, you know, he had his typical, I'm

Speaker:

moving to a different platform problems. And

Speaker:

in previous challenges, if

Speaker:

he ran to a wall, he was. And he's actually on the record saying this.

Speaker:

He said, when I ran to the wall, I just threw my hands up and

Speaker:

said, well, I'm just going back to Windows. He said, but now with all the

Speaker:

trouble that I'm having with Windows right now with security and

Speaker:

root level anti cheat and stuff like that, he knows what's going

Speaker:

on there. And he's like, now I am

Speaker:

motivated to put in the effort to migrate

Speaker:

to something else because this is.

Speaker:

This is just getting to be too much of a pain point. So.

Speaker:

Yeah, interesting. Yeah, here's, by the way, the page that you mentioned

Speaker:

on kde. So they do ship these around to be fair to people.

Speaker:

So sometimes we're at the very top, sometimes we're near the bottom

Speaker:

here. So I guess we. I mean, next time around

Speaker:

we'll be at the top again. So we'll see.

Speaker:

Maybe they're just trying to get more money from you. Who knows?

Speaker:

We're a big contributor. I think we're one of the. I

Speaker:

wouldn't say we're the largest contributor to kde, but I think we're

Speaker:

now moving towards the upper echelon because. Okay, we're a patreon. I'm just

Speaker:

joking. I'm just joking. I don't want to. I don't want to start any fights.

Speaker:

I don't want to start any trouble. Oh, no, no, no. The initiative that I

Speaker:

was talking about, though, is the bulletproof KDE initiative that.

Speaker:

Yeah, working title. That's where we're basically

Speaker:

freezing KDE at a specific version and then

Speaker:

we are pro providing them. And I won't get into all the

Speaker:

details, but we're setting up a continuous

Speaker:

integration environment and where we're able to

Speaker:

do automated tests within the data center. It's in

Speaker:

Finland. And then we work with an international team

Speaker:

for identifying. We manage the list

Speaker:

of key paper cuts and. And then

Speaker:

we pay them to help shepherd getting

Speaker:

those things fixed. And we work with a Kubuntu team, which we also provide

Speaker:

stipends for, to make sure that those get backported to

Speaker:

the release. So if you buy one of our systems, not only do we continually

Speaker:

test things like kernels and drivers before you

Speaker:

ever see it, to make sure that it always works. So we're kind of

Speaker:

like your remote IT department. Right? You know how, like, with your IT

Speaker:

department, you don't get an image until they verified it? Well, we

Speaker:

verify that we. Oh, that's pretty cool.

Speaker:

Yeah. So we actually, if you want to take a look at

Speaker:

what we verify, I'll share this tab instead. If you get a.

Speaker:

This is probably a good way to put it, but here let's just go to

Speaker:

support and you can look at the

Speaker:

system support and it walks

Speaker:

through everything that we do from hardware selection to

Speaker:

making sure. Like making sure that the dual ethernet

Speaker:

on the ZR Gen 2 which is coming up,

Speaker:

or the big beast we just looked at, making sure those work,

Speaker:

making sure three monitors work on sleep and

Speaker:

resume etc, we make sure that's there. We make sure the

Speaker:

kernel updates always work and that we have about 150

Speaker:

major KPCs that we go through. We make sure

Speaker:

that the Nvidia machine learning language, sorry,

Speaker:

machine learning libraries are available

Speaker:

and all these, which for non Linux people that's a

Speaker:

massive pain point. I will say making sure you get that

Speaker:

working like. No, that part of the reason why I

Speaker:

chose to get the Nvidia, the gtx, the DGX Spark.

Speaker:

Yep. Was that it was made by Nvidia and Nvidia

Speaker:

would be on the hook to supply those drivers.

Speaker:

So hold that thought because I want to. I want to come back to Nvidia

Speaker:

and Linux in a few minutes, but go ahead. Sorry. Well, it's all right. We

Speaker:

just. We spent the last. On the ZR Gen 1

Speaker:

and the M2 Gen 7. We

Speaker:

spent about 150 hours of development time on

Speaker:

creating and packaging everything for those

Speaker:

Nvidia drivers and the machine learning libraries.

Speaker:

It was a. That. That is not normally the amount that we

Speaker:

spend, but because we do support the machine

Speaker:

learning libraries and this was with Ubuntu

Speaker:

24.04. There were a lot of challenges

Speaker:

this time around. I think it's partially due to the newness

Speaker:

of the hardware and also due to the fact that there's more of an

Speaker:

emphasis on commercial hardware than they

Speaker:

used to have versus consumer hardware.

Speaker:

But in any event, setting all those things up, adding the

Speaker:

right repositories from Nvidia and then pinning to the right

Speaker:

machine learning and inference libraries is

Speaker:

a lot of work. And if you buy from us, you

Speaker:

outsource that. So we test that

Speaker:

before every kernel upgrade and we test that for every major

Speaker:

upgrade. For the Nvidias we actually have an

Speaker:

Nvidia meta package that we create that's validated for your

Speaker:

hardware. We also have a kernel which is spun

Speaker:

off of the mainline Ubuntu kernel, but

Speaker:

we will advance patch stuff that

Speaker:

we then upstream to the mainline kernel or to the Ubuntu

Speaker:

kernel that allows say proper hardware

Speaker:

integration or sound routing and stuff like that. So when you get

Speaker:

the system, you open it up, you turn it on

Speaker:

and everything works. And then it continues to work for the

Speaker:

entire support window. And the standard support window is three years.

Speaker:

So we keep bench models and we validate every

Speaker:

single model that's been released and last sold in the last three

Speaker:

years. So when you buy a system, you typically get three and a

Speaker:

half to four years of support

Speaker:

in basically IT team support for your system.

Speaker:

And then by the time you go off of that support, generally

Speaker:

everything is upstreamed anyway. So. But

Speaker:

it's. That's the reason why companies buy from us for

Speaker:

this purpose. So the other thing we do here,

Speaker:

and I'll change the sharing here because you'll. Is there's a lot of

Speaker:

integration that we do that is

Speaker:

people ask us are we a hardware company or a software company?

Speaker:

We're really a software company that does hardware because we have

Speaker:

to. Interesting.

Speaker:

Yeah. The best way to describe that, I would say is that when you buy

Speaker:

a Mac you get a full ecosystem

Speaker:

and you have a few key pieces of hardware

Speaker:

that they continuously integrate for. And so I

Speaker:

don't think anyone's going to argue that Macs are more reliable than Windows

Speaker:

systems because of the. And it's because of the limited hardware

Speaker:

and the fact that Apple does continue

Speaker:

for a number of years to do basically what we're doing, which

Speaker:

is they have a limited amount of hardware and they have

Speaker:

a limited amount of software. It's a vast support

Speaker:

surface, but it's manageable. And we're

Speaker:

doing the same thing for Linux is that that Apple is doing for their

Speaker:

proprietary os. But we have the advantage of being Linux native.

Speaker:

That is very valuable if you're deploying to Linux

Speaker:

as opposed to deploying to macOS. Yeah, that's

Speaker:

our philosophy. The other thing that you'll see and we're in

Speaker:

kind of transition here, so. And I'm actually gonna. This is the 26

Speaker:

out of 4 development. I'm in an alpha system. So if something

Speaker:

breaks here, don't go around and point say, hey look, they're

Speaker:

shipping broken stuff. This stuff isn't shipping for another. We're planning on

Speaker:

two to three weeks. Two weeks would be the very minimum. Probably three

Speaker:

weeks we'll be shipping once all the bugs are worked out. But

Speaker:

here's an example though of if you go to like support here,

Speaker:

we have all of these. Pretty much anything that you want to do

Speaker:

that's a workflow, but. And

Speaker:

so if you look at like Linux help, you can go to Chat

Speaker:

GDP and have it lie to you for six or seven iterations

Speaker:

on a question, or you can go to a forum

Speaker:

or a bug report and try to figure out how to get something done.

Speaker:

And we've all had the experience where it's kind of similar to Chat

Speaker:

gdp where it'll lie to you numerous times and hallucinate

Speaker:

until you kind of have to know where you're going to get the correct

Speaker:

answer. Or you could just go here and get a correct answer right off

Speaker:

the bat. Because it's tested for your system, so

Speaker:

it's tested for the OS for your system. So if you want to connect to

Speaker:

a phone, you can see that this is tested for a

Speaker:

previous operating system that's still supported. It's also verified

Speaker:

on the current one because we actually have people to go and test the

Speaker:

entire procedure on the OS

Speaker:

with the drivers that are in place. So

Speaker:

interesting. You can, you know, let's say I want to set up a

Speaker:

vpn. So you just go here, press VPN and

Speaker:

oh, look, I can connect to a VPN. And there you are,

Speaker:

VPNs everywhere. There's Taylor Swift. I was wondering when she was going

Speaker:

to show up. And this will actually walk you through setting

Speaker:

up a VPN completely for your system. And this

Speaker:

is a amalgamation of

Speaker:

over here running a Wireguard server

Speaker:

or installing OpenVPN and configuring everything

Speaker:

in an optimal fashion for your system

Speaker:

if you want to know how to do things right. And it's a

Speaker:

common workflow, we probably already have it here and

Speaker:

it's been tested for you. One of the things

Speaker:

that we do that our customers really like

Speaker:

is we make sure that they don't have Franken systems.

Speaker:

So you see that with a lot of people that are using Chat GDP

Speaker:

or using forms and stuff like that, they end up finding

Speaker:

suboptimal solutions after

Speaker:

maybe breaking their system a couple of times. Then they copy and

Speaker:

paste something to a configuration file somewhere and it works kind

Speaker:

of well enough and they limp along with it and they're afraid to touch it

Speaker:

because I might break it again. And then they have to reinstall the OS

Speaker:

for one reason or another or move to a different machine and they have no

Speaker:

idea what they did. If you contrast that to what we

Speaker:

do is that we end up finding those solutions

Speaker:

and we'll create them into this documentation. But

Speaker:

the way that we run the company is we actually push those

Speaker:

changes up into software so that you just

Speaker:

update your packages and. And those configurations are

Speaker:

done for you in a repeatable fashion, again, like an IT

Speaker:

department. So we even have

Speaker:

a pitch deck that actually shows how we do that. And

Speaker:

so our goal is always to move everything up the stack. Because

Speaker:

if you look at compliance, people doing the

Speaker:

right thing, if you put the stuff into the packages

Speaker:

that people update, their compliance is 100%.

Speaker:

If you put it in documentation, if you're lucky,

Speaker:

you might get 50% compliance or 30% compliance.

Speaker:

Right? And we're always pushing to get

Speaker:

100% compliance so that people don't have to worry about that.

Speaker:

One thing that's also interesting here is like, if you look at. Let's take

Speaker:

a look at image editing, for example, and we have a whole

Speaker:

subsystem that's in place here for,

Speaker:

we call Curated Apps. And this allows you to

Speaker:

like, say, start an application directly from the

Speaker:

documentation. And if the application isn't installed,

Speaker:

it'll actually add the cryptographic keys and

Speaker:

then add the repository and then download the program

Speaker:

and then install the program if necessary.

Speaker:

So you don't have to step out and do five or six different steps and

Speaker:

say, what was I doing? You just click on something and if

Speaker:

it's installed, it'll launch it like you see here.

Speaker:

And if it's not installed, it'll go all the way back as necessary.

Speaker:

It'll even encourage you to

Speaker:

even step you through, for example, installing,

Speaker:

say, Docker, which

Speaker:

requires service setups and adding

Speaker:

yourself to groups so that you can run as a regular user and stuff like

Speaker:

that. It'll actually walk you right through that.

Speaker:

So if I go to support, let's say, let's go to Curated

Speaker:

Apps right here. Now, I haven't checked this out on here.

Speaker:

We're still in alpha, but I believe this is actually already in

Speaker:

place. So let's take a look for Docker. So click

Speaker:

on Docker and there it's not installed. So do I want to

Speaker:

install it? Yes. Now I'm asked for my password

Speaker:

and away we go. And now it's actually downloading the system. It's

Speaker:

added the repository for the official Docker release.

Speaker:

It's added the cryptographic keys for the actual Docker release.

Speaker:

It's installing it right now. And then it's going to guide me through changing

Speaker:

the groups again. We're just really

Speaker:

focused on making sure that people are

Speaker:

prepped for success. Here you go. It

Speaker:

says Dockers installed, but the user dev must be added to the Docker group to

Speaker:

run as normal user. Do you want to add to the group. Now

Speaker:

say yes, now I have to do that as

Speaker:

root. Now it's been added,

Speaker:

Docker's installed. But you got to log out and log back in if you want

Speaker:

to use it as a user.

Speaker:

It's fantastic. That's amazing because this does. I know a lot of people that

Speaker:

are very, very lost right now in terms

Speaker:

of Windows isn't what it used to be. And

Speaker:

there's a lot of people that are looking for a new place to land with

Speaker:

high performance hardware. And I can only assume that you

Speaker:

could probably run this as the base operating system and kick off as many

Speaker:

VMs, particularly if you have enough RAM, kick off as many VMs as

Speaker:

you need for Windows. Right. If you really absolutely needed a Windows like

Speaker:

a pure Windows machine. Oh, absolutely. In fact, if you.

Speaker:

That sounds like a common workflow. Let's see if we have something that fits that

Speaker:

category. Oh, here's how you stall

Speaker:

virtual 11. Right. And by the way, all of these right

Speaker:

here have, are available

Speaker:

in blow up format, if you will.

Speaker:

So we can just walk right through these and it'll just go through all the

Speaker:

steps for you. This has been validated for Windows 11

Speaker:

on the latest OS. You can see over here that it's already applied

Speaker:

to this os. I mean this is brilliant because

Speaker:

there's so many times when there's no instruction manual, there's no

Speaker:

instructions anywhere. You kind of have to just,

Speaker:

you know, Svengali it yourself. But to have these guided solutions

Speaker:

is just. Is really, I think

Speaker:

reduce anxiety. Especially with Linux.

Speaker:

Right? Linux, you're kind of, you know, good luck

Speaker:

is kind of like the. You're kind of

Speaker:

left to your own devices and your own skills, which for some people is perfectly

Speaker:

fine. But not for everyone. Oh yeah, and even

Speaker:

the people that. There's a bit of a dunning Kruger effect out there.

Speaker:

Yes, smidgen of that. Especially people that get behind

Speaker:

chat gtp. Right. And you know, I used it just the other day and

Speaker:

I was trying to solve a problem where I knew already what the answer was.

Speaker:

It was actually the answer was

Speaker:

looking to pull down the latest Jetbrains toolbox

Speaker:

and install it. So one of the curated apps here that

Speaker:

is not finished for this version of the release is

Speaker:

installing the Jetbrains toolbox. So I knew that I had to go out and

Speaker:

pull a web API and make a request to

Speaker:

get the current version and then build a

Speaker:

URL to create a download poll.

Speaker:

Being a developer for three or four years at

Speaker:

least. Yeah, multiply that by a

Speaker:

bit. But I knew that that's where we needed to be. Right.

Speaker:

And so I went and used GTP and asked,

Speaker:

how do I do this? And I had to ask literally six times.

Speaker:

And it kept giving me the wrong URL to go and make the request on

Speaker:

it. And I was like, you know, I could have just looked up for the

Speaker:

Jetbrains API and I would have gotten there in

Speaker:

a search and then a quick query. But the

Speaker:

problem with those answers were they were all

Speaker:

extremely confident. Oh, yeah, you just need to

Speaker:

do this and this and this here, cut and paste this. It won't hurt.

Speaker:

We're seeing an uptick in people breaking their systems because they're using chat

Speaker:

gdp. Why doesn't that surprise me? Yeah,

Speaker:

we're actually been talking about creating a local

Speaker:

LLM that will basically consume all of these

Speaker:

documents that you see and use these as the primary source

Speaker:

instead of going out to the web and scraping that. And

Speaker:

that way people will have an interface that will be available

Speaker:

directly from their toolbar and

Speaker:

that they're. They will be just as confident. But

Speaker:

unlike the confidence of chat gdp, most

Speaker:

of the time, in my experience, it'll be right.

Speaker:

Bonus. Another thing here is you

Speaker:

can see that we have service guides for all of our systems that.

Speaker:

And we have the same structure for every one of them, including what the bio

Speaker:

settings should be for any one of the systems,

Speaker:

so. Oh, very cool. Yeah, you click on the show you how to swap out

Speaker:

your ram and we don't void the

Speaker:

warranty if somebody opens up their case, so.

Speaker:

Interesting. Yeah, I.

Speaker:

Where can folks find out more and get their hands on one of these sweet

Speaker:

systems? Well, I just go to our website, which is

Speaker:

the. There's our. That's the.

Speaker:

That's the new interface you're seeing there. I shouldn't say

Speaker:

new. That's an overview. So we have different

Speaker:

things here. Oh, very cool. Yeah, so

Speaker:

it's a beautiful interface. It's just.

Speaker:

Kfocus.org is our website. Okay. We used to have

Speaker:

kfocus. Com, but we had somebody steal it

Speaker:

from us and to give it back. Yeah, I

Speaker:

won't go into all the details on that one, but watch your backs, people.

Speaker:

Yeah, just go here. And I think it's pretty straightforward.

Speaker:

And yeah, we'll be. I

Speaker:

wouldn't recommend buying it right now because in the next couple of days.

Speaker:

So if you're watching this in the next couple of days, then yeah, buy right

Speaker:

now. But we'll be updating all these

Speaker:

models to 26.04 and we don't expect to

Speaker:

release any or ship any other models with the

Speaker:

earlier version of the os. So we're basically on a little bit of a

Speaker:

pause as we finalize that and finalize the validation

Speaker:

for all the new hardware. It's a big hard push right now that we're

Speaker:

going through, but it's going to be pretty awesome.

Speaker:

It's our again, Candace was there at the

Speaker:

beginning from 2019 and

Speaker:

this will be our fourth major or

Speaker:

fifth major OS upgrade. And fourth actually.

Speaker:

Fourth and

Speaker:

yeah, it's better than ever. Well, there's one thing that I do want to show

Speaker:

you that I think that you would very much appreciate

Speaker:

and let me put it here. One of the things you run into

Speaker:

with immutable systems. Immutable systems are great

Speaker:

until you don't want them to be immutable anymore and then you're stuck.

Speaker:

We put about 1200 engineering hours

Speaker:

into a thing called

Speaker:

rollback. Oh, very

Speaker:

cool. So you can, it's. That could be very

Speaker:

useful. What system rollback

Speaker:

does is it takes a snapshot of your root system and

Speaker:

boot system and then it is immutable.

Speaker:

People that are creating immutable file systems or immutable

Speaker:

systems are basically doing the same sort of thing.

Speaker:

But here you're doing it at your

Speaker:

leisure or you can have it done automatically. You

Speaker:

can see up here we have automatic snapshots on here and I'd give you

Speaker:

a demo on it, but it's not on this system because it's a development system.

Speaker:

But there's people that you'll see post recently there was

Speaker:

on Twitter where one of the guys said the best system you can get is

Speaker:

Kubuntu Focus because they're amazing rollback system.

Speaker:

And this is what he was referring to. This allows you to take

Speaker:

snapshots, roll them back and every time you upgrade your system

Speaker:

it takes a snapshot before the upgrade. So it

Speaker:

causes problems. You can roll back. So not only are we pre

Speaker:

checking all the major updates, like all the kernel updates etc and all

Speaker:

the Nvidia driver updates, but even if a minor

Speaker:

update causes a problem, let's say an app update, you can just go

Speaker:

over here and roll back to the previous one version. That's

Speaker:

fantastic. We also wrote an event driven

Speaker:

monitor. It's a throttled and

Speaker:

debounced event driven system.

Speaker:

So it doesn't have a major

Speaker:

impact in performance. It's a very tiny impact of performance.

Speaker:

But if you end up pushing this copy on

Speaker:

write file system to the point where you don't have enough space,

Speaker:

it will actually pop up and warn you immediately and

Speaker:

it will also turn off the automatic snapshots until you

Speaker:

fix it. And it also guides you on how does turn off the

Speaker:

snapshots until it's fixed. Interesting. Yeah, that was a

Speaker:

part of the reason why it was 1200 hours is because we have the

Speaker:

whole solution. Also, unlike a lot of

Speaker:

software, this was accelerated tested. I come from

Speaker:

automotive design background so like when if we screwed up,

Speaker:

people died and you lose your job. We use a lot of

Speaker:

spec driven development techniques that have been around

Speaker:

for probably 50 years or longer in automotive that have been

Speaker:

proven out. And so we use design for

Speaker:

design failure mode effect analysis with RPMs

Speaker:

and KPCs and then we also use accelerated

Speaker:

testing. So we actually had laptops set up for this where

Speaker:

they ran through multiple three year cycles

Speaker:

of taking snapshots, installing packages, rolling

Speaker:

back, taking more snapshots and that was all automated. So we just had

Speaker:

systems that were like 125 watt light bulbs burning up

Speaker:

constantly doing all of that. But we wanted to make sure that it

Speaker:

was reliable for users. And so. And then we

Speaker:

also did the same thing like with. Or we do a

Speaker:

related thing which is make sure the BTRFS file system that

Speaker:

this is based on has appropriate performance and we tune that

Speaker:

as well. Basically it's like again it's like you're going into a big

Speaker:

company, you've got an IT department that looks after you

Speaker:

and make sure that everything just works and keeps working.

Speaker:

That's cool. Yeah. All you got to do is update. Awesome.

Speaker:

We'll make sure that we put the shit linked here and you got me,

Speaker:

got me. You got my attention. Good. So very

Speaker:

cool. And I think it's a good alternative for people that.

Speaker:

I think it's a great alternative for people who want to get into Linux but

Speaker:

maybe don't are afraid of it. And all these things you put in in place,

Speaker:

I think remove a lot of that fear and I think that's great.

Links

Chapters

Video

More from YouTube