My Move to SUSE
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 57 | |
| Author | ||
| Contributors | ||
| License | CC Attribution 3.0 Unported: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
| Identifiers | 10.5446/54468 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
|
openSUSE Conference 201727 / 57
3
6
10
15
16
19
21
23
24
27
28
29
34
36
38
40
41
42
46
47
51
53
55
00:00
Distribution (mathematics)Computer animationLecture/ConferenceMeeting/Interview
01:17
EmailPhysical systemDependent and independent variablesDistribution (mathematics)Integrated development environmentSoftware testingEnterprise architectureXMLComputer animation
02:34
DichotomyServer (computing)FingerprintData centerFormal languageDistribution (mathematics)Server (computing)LaptopData centerMoving averagePoint (geometry)FingerprintWebsiteOperating systemEmailFreewareOperator (mathematics)Computer animationMeeting/Interview
05:41
Server (computing)Programming languageMultiplicationLecture/ConferenceMeeting/Interview
06:10
DichotomyServer (computing)FingerprintData centerFormal languageOperating systemPoint (geometry)PrototypeHand fanComputer programmingJava appletComputer animation
06:51
Point (geometry)Goodness of fitRight angleLecture/ConferenceMeeting/Interview
07:14
Moving averageEnterprise architectureState transition systemSystem of linear equationsOperating systemInstance (computer science)Distribution (mathematics)Goodness of fitOpen setPhysical systemIntegrated development environmentRight angleOffice suiteDecision theoryType theoryServer (computing)Level (video gaming)Point cloudCycle (graph theory)Multiplication signMechanism designLaptopEnterprise architectureBuildingSoftwareLocal ringBlogModulare ProgrammierungMetropolitan area networkPoint (geometry)Installation artPerspective (visual)Data centerProcess (computing)ReiserFSSlide ruleCodeFrustrationBitComputer animation
12:39
FingerprintSoftware testingPhysical systemData recoveryDefault (computer science)Decision theoryOpen setBitFingerprintCurvaturePhysical systemMultiplication signCommitment schemeMoving averageProgram slicingSoftwareLevel (video gaming)Confidence intervalRollback (data management)Existential quantificationDecision theoryNumberBooting1 (number)QuicksortConnectivity (graph theory)Extension (kinesiology)EmailBlogConfiguration spaceProduct (business)Configuration managementDefault (computer science)DampingRight angleFlagServer (computing)Operating systemDistribution (mathematics)Software testingVarianceClassical physicsLoginMereologyComputer configurationDirectory serviceProjective planeRepository (publishing)LaptopPoint (geometry)Open setLecture/ConferenceComputer animation
22:17
Process (computing)MereologyCycle (graph theory)Right angleOpen sourceHyperbolischer RaumArithmetic meanMeeting/Interview
22:51
Open setDecision theorySoftwareMereologyDistribution (mathematics)PlastikkarteIntegrated development environmentOperating systemCurveArithmetic meanOpen sourceData managementBusiness modelMultiplication signEnterprise architectureReflection (mathematics)Server (computing)Open setPhysical systemCore dumpStability theoryOnline helpComputer animation
25:53
WeightVideo gameDistribution (mathematics)Parameter (computer programming)Computer animationLecture/Conference
26:30
WeightChainData managementPhysical systemUser interfaceLine (geometry)Disk read-and-write headLevel (video gaming)Integrated development environmentAreaSoftwareDifferent (Kate Ryan album)Point (geometry)Negative numberSoftware developerBuildingRight angleComputer animation
29:08
Event horizonCartesian closed categoryHypermediaLecture/ConferenceComputer animation
Transcript: English(auto-generated)
00:09
All right, now I'm going to take advantage of your hospitality and I'm gonna give you a softball talk. So my name's Thomas Hatch, I'm the CTO of SaltStack. And so I'm giving a keynote tomorrow.
00:23
I usually have to give a lot of very technical talks and very in-depth and visionary talks. And I never get to give a talk for fun. And so I presented, or sorry, I submitted a softball talk.
00:41
And probably just because you're being nice to me, you let me get up here and talk about it. Basically what I wanna go through is how I introduce modern SUSE to people. Because I have some strange looks, I will admit.
01:02
When people ask me about Linux distribution, I'm running and I go, well, SUSE, of course. So let me kinda just talk through some of these things. And I noticed that there are other talks that use a Dr. Strangelove reference in the title. If you haven't seen the movie Dr. Strangelove,
01:21
it is absolutely hilarious. Okay, so it all started with a guy named Bo. So I get this email from Bo almost two years ago. And he says to me, SUSE is going to start using
01:43
the ever-loving crap out of salt. And my response was, so? You're SUSE, I don't care. That's honestly what I thought. And so a little later, honestly only a few weeks later,
02:04
what I ran into was that we needed to reinstall the base operating systems on one of our test environments in our QA lab. And I thought, well, I have to at least look at SUSE and I have been deeply offended
02:21
by a certain other enterprise Linux distribution lately. And so I started to look at SUSE. Oh, I have a picture of Bo. If any of you don't know Bo, this is what he looks like.
02:40
Okay, that's actually how I honestly describe him to people, big nose. All right, so I had to put a couple of my old SUSE biases aside. And a lot of this came from, honestly, one of my biggest old SUSE biases
03:01
is that I felt like you did set up the Apache configs in a weird way. That bothered me. I've gotten over that. And so this is also, this is the main thing that I run into when I start talking to people about SUSE.
03:20
Again, I'm from the United States where things are kind of crazy over there right now. And so I'll admit SUSE is not a very popular distribution in the US. And so I usually get this kind of kickback, ironically, even though I'm in Utah and you guys have some ties over there.
03:43
For better or for worse. Okay, and I will have to admit on the third point, I kind of like Yast now. I'm almost embarrassed to say it. It's really convenient.
04:04
Okay, so the main thing that started to change my mind was how OpenSUSE changed their approach to releases. Again, right after Bo sent me this email,
04:23
I go to the OpenSUSE website and go, okay, I'm gonna take a look at OpenSUSE. And I was unaware of the whole leap tumbleweed thing that was going on. And the big problem that I generally had was,
04:42
I'm an old Arch Linux guy. I love Arch Linux. I used to be an Arch Linux packager, et cetera, et cetera. But in the data center, I would always use CentOS. I wouldn't use anything other than CentOS or anything related to CentOS, just CentOS.
05:01
And I found it frustrating that I lived in this two distribution world, where I felt that I needed a completely different distribution for my laptop than I needed for my servers, whether they were in the data center or at my house. And upon learning about Leap, I got really excited
05:22
about this, because all of a sudden, I can use a rolling distro, because six month distribution releases are silly and dumb. So, hooray, I can use a rolling release. But also, I've got a completely free operating system,
05:43
which is more than sufficiently stable to run the vast majority of my server needs. And so, I like to compare this to using multiple programming languages. We already had a little trolling from Joe earlier, if you were in this room,
06:01
about using multiple programming languages and how that's morally wrong. And I agree with other commenters that the solution is not more Perl, it's less Perl. I'm not a big Perl fan, or Java, sorry. But then again, I would endorse the use of Julia,
06:23
if anyone's used Julia, it's brilliant. I'm not serious. But in programming, often we do have to say, we're gonna use something like Python for high level, get a lot of stuff done, prototyping, and then have C so that we can optimize performance
06:42
when we need to. And I didn't like having the same problem inside of my operating systems. And that's a really strong selling point for people to be able to explain how well tested Tumbleweed is. Tumbleweed is beautiful.
07:01
Is Richard Brown in here? Oh, that's good. Okay, yeah, Tumbleweed is beautiful. All right. And so, I begin to feel that this idea, this concept of a rolling release,
07:20
plus an enterprise release, and the overall way that SUSE develops software, develops the operating system and pushes it out was far and away the optimal mechanism. And again, I felt that this six month release cycle concept is, well, I think it's a waste of a lot of people's time.
07:43
And then the benefit I felt is that the enterprise software existed in the right place. That it was the right type of upsell and the right type of ramp to get somebody moving from certain environments on a rolling release, certain environments on LEAP, certain environments on SLEZ.
08:01
And I felt that was a lot smoother. Now, I don't remember who wrote this blog post. But someone from SUSE wrote a blog post about how SUSE makes decisions and then commits to them. Was that you, Joe?
08:20
I really didn't mean to compliment Joe. Actually, it was a really good blog post. And this is something that really struck me because coming from certain other distributions whose names may or may not appear on this slide,
08:42
I'd become very frustrated with things kind of flopping around. You know, should I be using Xen? Should I be using KVM? Is ButterFS still gonna work in 6.9, et cetera, et cetera? And I began to really appreciate SUSE and the fact that they still put up
09:01
with RiserFS, for instance. Good job, guys. I think RiserFS was very brilliant for its time. Obviously not quite brilliant enough. Now, all right. So the next major problem that we're running that I feel is very important from the perspective,
09:22
particularly deploying inside of a data center, has to do with the ability to backport software packages. So let me go back to the story. We're at Solstack, we're determining what distribution to use underneath a new cloud deployment. And it was a pretty small cloud. And so we go, well, we'll just use Open Nebula,
09:41
get this thing up and going pretty quick. Doesn't need to be complicated or fancy. And we start to install Open Nebula on SUSE. And I told one of my engineers, okay, we're gonna try SUSE. And it was Tuesday. And I said, if it comes to Friday
10:02
and you're frustrated and mad, we can back out. Because no one in the company had ever used SUSE at this point. This engineer is named CR, wonderful, wonderful man. He comes back to me on Thursday
10:21
and I'm fully expecting him to say, yeah, Tom, that's the SUSE garbage. And he comes and he walks into my office and he starts talking to me about something completely different, right? And I ask him, hold on, hold on, hold on. What about the server deployment? I mean, weren't you doing this? Why are you asking me about code?
10:41
And he says, oh, no, I'm done. So everything's okay? Oh yeah, everything's fine. I'm like, so did you use SUSE? He says, yeah, yeah, it was really interesting because at that point, again, it had been two days. He's in my office and he says,
11:01
not only have I reinstalled all the servers, but I mean, there's no more Ubuntu on my laptop. I'm running Tumbleweed. I have been converted. And the thing that did it for him was OBS. And he stayed for Snapper, but more on that later.
11:21
But the thing that did it for him was OBS because he started to deploy Open Nebula and the Open Nebula packages he could find for OpenSUSE were horribly out of date. And this is a showstopper a lot of the time that it becomes very, very difficult to say,
11:41
ah, nuts, I really wanted to use this particular piece of software, but just because the vendor doesn't happen to be all over it, I'm going to have to go through the trouble of building my own packages. But you guys have OBS, thank you.
12:00
I don't have to build a local package, build environment that shoots stuff, thank you. And so he's in my office and he says, oh, I just tweaked a couple of existing RPMs, put them on OBS, and we were done. They built in a few minutes, and we were able to deploy our cloud.
12:23
So thank you for OBS, thank you for being brilliant into building that. I'm not actually gonna badmouth anything else from this stage apart from a certain, a certain other build system fills me with rage. Okay, finally talking a little bit more about Tumbleweed.
12:47
There's an old joke in the Arch Linux community. Has anyone in here ever run Arch Linux before? Okay, okay, I'm going to assume the number's actually higher and that most of you are just afraid to admit it in this room.
13:05
So the old joke in the Arch community is something's broken, so I'll just update in a few days and that'll go away. Sure, something else will be broken, but at least this problem will go away. The Arch isn't only a rolling release,
13:21
and don't get me wrong, I still adore Arch. Arch isn't only a rolling release, but it's kind of like a wheel that every time it turns, it's flat in a different place. And I was astonished by how thoroughly
13:40
Tumbleweed gets tested. And as an Arch packager, I was very aware of the amount of effort that certain Arch packagers put into their packages. Wasn't always quite to the same level as the Tumbleweed project, to put it as kindly as I possibly can.
14:04
Sometimes it was rather reckless. One of the main reasons I love packaging salt on Arch Linux is because if there's ever an upstream issue with a dependency in any plausible or conceivable way, we will find out about it in Arch very quickly.
14:25
All right, so the ridiculously thorough amount of testing is something that I'm very, very grateful for. So in the old days,
14:47
when you would install a Unix-style operating system. And if you install FreeBSD, who here has installed FreeBSD in the last year?
15:01
Ooh, very nice. I haven't installed it in the last year, so I'm running on like two years ago on this reference. So I apologize if I'm wrong. Don't get after me. Okay, so if you install FreeBSD, you can still install it with slices or partitioning it in the classical Unix way, right?
15:24
You've got your var subdirectories isolated, et cetera, et cetera, so that if something logs like crazy, everything's fine, right? One of the things that blew me away just installing SUSE was how meticulously well
15:44
the ButterFS deployments are laid out. That they are very carefully thought through, that they are by default matched with the packaging and deployment and configuration options, and how they line up the right copy-on-write style,
16:04
et cetera, et cetera. All of the right flags are there for all of the right ButterFS components. And I was just very impressed by that. And then the next thing that I found astonishing,
16:20
this engineer, CR, I told you about, came up to me a couple of months after we deployed these systems and asked me if I had ever used Snapper. At that point, I hadn't heard of Snapper yet. After I was introduced to Snapper, I actually sent, I think I sent Bo an email saying, so I've learned about Snapper and OpenQA
16:41
and OBS, what other magical SUSE things am I missing? And he introduced me to quite a few more, like OSC. I love the OSC command line, by the way. I don't really know that those guys would ever need to rewrite that.
17:04
Sorry, for those of you who weren't here, that's what the last talk was about. Okay, I became instantly infatuated with Snapper. I mean, this thing is cool.
17:21
To the extent that I started purposefully doing all sorts of crazy things to break my laptop, because all of a sudden I felt far more liberated than I had ever been before. And then thoroughly enjoyed just booting into an old snapshot. It was excellent.
17:40
And all my problems went away. I could install all of those third-party repositories, you know, the ones that you install but you don't tell anybody about, especially Richard Brown, because he gets all upset. He's still not here, right? Yeah, and then the abilities
18:01
just boot into the past so seamlessly. And so similarly, I was thoroughly impressed by the fact that Snapper ran automatically, right, whenever you ran Zipper or Yast. And that I built Snapper support
18:20
directly into Salt's configuration management runtime so that every time you run Salt, you can snapshot back. As the first time, and still when we get customers who come to us and they say, well, we want, we want full rollback. And we explain to them that that's a myth.
18:42
Unless you install SUSE. Okay, oh, I already mentioned this because Joe brought it up in his excellent blog post. But SUSE sticks it out
19:00
and they seem to have very good judgment because they chose Salt. I'm not biased. Okay, but yeah, I did mention this before, but SUSE sticks it out. My dad, back in the late 70s,
19:24
spent a few years living in Germany. And so he would tell us stories growing up about the Germans and how they are different and fantastic in very specific ways.
19:40
And one of the things that he would say is that Germans will take a long time to make a decision. Because they want to make the right decision because they don't want to have to back out of that decision. Now, I don't know how true that is. I'm generally under the impression that it's true, interfacing with you lot. You seem to take careful, wise decisions
20:01
and then stick with them. But again, this is something I've deeply grown to appreciate in the way SUSE is engineered and also in the commitments that SUSE makes to its customers and its users. I feel much safer over the long run with SUSE
20:21
than I do with other operating systems. Because, I mean, I know that you guys are gonna make mistakes, that happens. And I'm very forgiving of those mistakes. But having that long-term confidence is something that I really like.
20:42
The fact that I feel very comfortable that if SUSE says you should use Software X, that they're not gonna back out and change that decision on me really soon. Yeah, and thank you for not using upstart.
21:00
I deeply appreciate that. Have you ever tried to automate upstart? Okay, so in a nutshell, this is my pitch. When people say, what distro are you using? And I say, SUSE, duh.
21:21
And they give me that funny look like I'm a madman. And they want me to like use some things. It blows my mind because they say, what Linux distribution are you using? And I say, I'm using SUSE, of course. And they give me this look like, seriously, like I'm crazy. And then I ask them what Linux distribution are you using?
21:42
And they're like, Linux from scratch. And it's, who's the crazy guy? Or they say something like, we deploy production servers on Fedora. No, I actually worked for a company that did that.
22:03
And it worked out great all the time. It was the government, the US government. They were brilliant. Actually, it was part of the US intelligence community.
22:22
Ironically, the side you don't hear about because they do their job right. Okay, so SUSE has the best release policies and release cycles. I'm not, I feel like I can back that up with more than just opinion and hyperbole.
22:42
The world of open source is a rolling world. Not having a rolling release that is stabilized means that your users must always be behind the curve. And Tumbleweed solves it in a way
23:01
which is infinitely more elegant than any other rolling release. Having an open source release which is stable enough to run in a server environment, I believe is an extremely important aspect of a Linux distribution.
23:20
And deploying that piece of software in such a way that enables users to get to know SUSE and get to know what an enterprise and an extremely stable SUSE environment feels like is a smart business model to help drive revenues. And that's another thing that impressed me about SUSE
23:41
is that unlike all but one other Linux distribution, you guys are profitable and growing. And I've become increasingly wary of trusting open source software which seems to be on a revenue dead end. And a lot of it is, a lot of it is.
24:02
Certain pieces of software which have an extremely high amount of hype and have raised hundreds of millions of dollars make no money still. And it's terrifying and it's making it hard to raise money. It's getting easier now for us because we make money, it helps.
24:22
Okay, so OBS is brilliant. And again, that's a huge part of that pitch is to emphasize that, so when I introduce Linux operating systems to somebody I say, look, the most important part of a Linux operating system is its packages
24:41
and its package manager. That is the core of the operating system. And how well those packages and its package manager are managed is a direct reflection on the quality of the distribution and the quality of the release. And you guys are doing that right.
25:01
It is not a cumbersome mess unless, well, Joe was explaining that it's got Perl in the background, but whatever. I can overlook this because I endorsed Julia earlier.
25:23
Let's see. Open QA is amazing. Your packages are brilliant. Your release policies don't suck. You solve the two distribution problem. You guys stick to your guns. And it took me a long time to admit that Pac-Man
25:42
was no longer the world's best package manager. I think it's Zipper now. So in a nutshell, thank you for making such a fantastic Linux distribution
26:04
and for making my life using Linux easier. And Richard Brown is absolutely fantastic. All right, any questions, comments, arguments, rebuttals? If anyone honestly wants to argue with me,
26:22
then that would be great since I just told you all you're the best. Yes. What feature requests do you have? So as much as you love what OpenSUSE is doing and what SUSE guys are doing, helping out, et cetera,
26:44
what's the itch that you would love scratched? In all honesty, I'm not sure. Mostly because you take such darn good care of my software that I write.
27:07
We could probably see a few more IDEs packaged. But I use K develop. So I know my credibility just went down.
27:22
That's German, right? Okay, any others? Yes. Yeah, first of all, thanks for this brilliant talk. I'm sorry I have to rain on your parade. I can agree to many points and maybe it's just because I'm still on my journey and only halfway there.
27:40
I have a red head fedora background as well. So I joined SUSE last year. You are already one year ahead of me basically. You have two years of SUSE experience, right? About one. Okay, okay. I will admit when I introduce SUSE to people, I do introduce it to them and say something along the lines of SUSE's brilliant
28:01
as of about two years ago. Okay, so the only thing that I cannot agree on is build systems because basically all build systems suck. There's just some build systems that suck less than others. And my question is what sucks so hard about Koji?
28:22
So in the defense of Koji, I haven't used it extensively for a few years. Mostly it was user interface being very, very cumbersome. And also their entire build chain and package management chain is spread across a lot of different areas.
28:40
But again, I haven't done a lot of work with the Red Hat build system in a few years. So that's also why I probably shouldn't have said anything negative about Koji. Setting it up is a nightmare, but I'm sure setting up OBS isn't a walk in the park either.
29:02
Okay, I will happily rescind the stage.