The Godfather of DevOps offers his advice for those getting started

Earlier this year, at DevOps days Austin 2022, I sat down with the godfather of DevOps, Patrick Debois.  I asked Patrick what advice he would offer those just getting started in DevOps or those early in their journey.  The three things Patrick focused on were the importance of finding the next bottleneck, the need to reframe your thinking and learning to loosen your control.

Transcript -- Patrick Debois - DevOps wisdom  – DevOps days Austin 2022

Barton: All right. Coming to you live from DevOps days, 2020, uh 2022. Oh my gosh. That's right. We've been on hiatus for COVID for two years and I don't even know what it is. So I'm sitting here with the godfather of DevOps, Patrick Debois, Patrick, how are you doing today?

Patrick Debois I'm great.

Barton So you've seen this DevOps things from, from ground zero.  [00:00:20] I someone were starting out in DevOps today, what would be the advice that you'd give and what are some of the things that they should be keeping in the back of their minds, as far as goals, things to look out for, et cetera.  

Patrick Debois: Yeah. It's a good question, people have asked me how to get started, how when they got started, how to move on and then when they’ve moved on, what to do next  [00:00:44] My personal answer is pretty simple. I'm a system thinker and I always say, find the next bottleneck. It could be that your bottleneck somewhere [00:01:00] regarding wanting to get faster to production so you can improve either on your deployment scripts or your servers or your code, or it could be that I'm paged every night and I need to solve something about my monitoring and they're all great things.  [00:01:16] , and then while you are trying one, solving one bottleneck to another, I think what you should be aware of is that once you do something, your bottleneck might change. So don't go all in like I'm gonna spend three years fixing my monitoring system, three years on doing my automation.

[00:01:37] So I made a comparison about a barrel of water.  if you wanna stop the leaking of the water It doesn't make sense to build one side completely up. It's going around and every time reevaluating, what's my bottle neck.  [00:02:00] You might at certain points think, okay all this is  what people think about  automation monitoring and so on sometimes.

[00:02:09] And it's something Ive learned, sometimes your bottleneck is outside of your control. It could be in the hiring. It could be the way people are compensated and it creates another dynamic.  It might be the way sales and marketing is pitching your product, that they can't sell it. So there's more money.  [00:02:30] So you can't hire people and they don't get the good money. So you see  it's a chain of things.

Barton: Which I think gets back to your point, which you've talked about before, which is the idea of putting it in the context of the greater business goal.

Patrick Debois: Yes, exactly.

Barton And what tends to happen is what you’ve call “technical gravity,” meaning we always want to go back to the tools.

Patrick Debois: [00:02:51] Yeah. Cause it's easy to understand.

Barton So how would you recommend that people keep their mind [00:03:00] focused on the goal as it were as opposed to the tools?

Patrick Debois:. Well, again, coming back to what's your bottle neck.  If you apply a new tool or you want to try something out, ask why am I doing this?  What am I improving with this? And it's okay if you just say the improvement is just that I wanna learn a new tool, which is totally valid.  And the answer might also be, well this is a cool new but it's not improving things that much. And I need to s something else.

[00:03:40] So there's this concept: boring tools. They just work. Right. And it's all great. So you don't need to do what people call rese driven development work.  You're always doing that kind of having to do the next big thing.  There's nothing wrong about learning the next [00:04:00] big thing, but the whole concept is you need to build up a kind of a tool chain, right.

[00:04:08] About different things you need to apply in different situations. And that's kind of the learning curve. One, how you go through the tools, when to apply them, when it's useful, when it has effect. And that's kind of how I think about it. I already mentioned that your bottleneck might change and that's not the only thing that would change, the industry might change, the way people think about problems will change.

[00:04:32] Over the years, people have different concepts, the complexity has increased, and then your old solutions might not be the best fit for the job and that's kind of a continuous rat race, , okay, are we still doing the right thing? When people moved to cloud, they were applying the concept of traditional virtual machines on their cluster.

Barton: It’s like when railroad tracks were put in and some people put their horse and buggy on the railroad tracks/

Patrick Debois: [00:05:00] Exactly, exactly.  So it's kind of reframing, rethinking.  I know a lot of people say, oh, it's not about the tools, but me personally, I believe there are cases where sometimes a new tool can help you bootstrap new behavior if you're doing it right.  Sometimes it has the opposite effect, obviously, but there's definitely a relation. Unfortunately in our industry, it will never end. Aalthough if you have gray hair like me, you go in circle, it's , oh, aren't we doing the cloud thing and getting back to the mainframes?

Barton: back to the 3270.

Patrick Debois: Yes exactly. And yeah. Okay that's me being old but there's shuffling around of complexity in many cases. I think it's one of the challenges we have right now, tools have exploded new tools, more tools. And we've been in the cloud world and cloud native Kubernetes came around and, and that's what kind of [00:06:00] makes it interesting is that we keep, we keep having to challenge ourselves.  Move away from our beliefs, that what we did five years ago is still the right way to go.  It's easy for technology to keep pulling us in. It's, yeah, we finally managed to get this tool working right. Another migration.

Barton: Well, then you get technical debt because you buy all these new toys and when Christmas comes, you get more toys  

Patrick Debois: But it doesn't care what programming language. It doesn't care what tool,  If you're passionate, you're hitting the right problem with your tool by all means, go for it. And I know it's generic. Its like a consultant who says it depends.  But it does and nobody's gonna do the thinking for you.  You know your situation best, your contacts best and don't let anybody fool you with the next tool, but please look at it, be critical and see how that works for you.

Barton:  Yeah and I think what you said too about marketing and sales and alignment because it doesn't help if you're doing this great thing and then they're disconnected.  You're disconnected from the business because you're working at cross purposes. So if you were to pick one thing, that's something to look out for in the future of DevOps, some new way of doing things, some new technology, some new way of thinking of it. Is there anything that's that has caught your eye?

Patrick Debois: I think we're still trying to be in control and we're still treating it as if we're the masters of puppet that control what goes in the system and so on.  I think if you transpose this to nature, trying to  frantically control the [00:08:00] environment is probably not the right way to scale, right.

[00:08:04] At a certain point, you need to have a way that that system will do parts of this. And you become more of architect or kind of the person who is escaping the system rather than having to run the system itself. So you're kind of creating opportunities, you're creating the environment and I think that's probably gonna make us a little bit more sustainable.  I know there's auto scaling, but, but still, we want to control how it does, how the scaling works

[00:08:40] Probably emerging behavior that we're, we're tapping into and it feels like I have to give up control. Yes. And it's the same thing with autonomous systems, autonomous teams, you don't control.

Barton: They say, if you love something, let it go, right.

Patrick Debois: I'm a father. Don't try telling kids what to do,

Barton, especially teenagers.

Patrick Debois:  Absolutely. They don't listen, but they'll learn themselves and you create a safe environment for them to learn that.

Barton excellent. Patrick Debois, a pleasure is always, thank you,

Patrick Debois:  you too. Thanks.