DevOps Workshop: Modern Technical Practices
DevOps Workshop: Modern Technical Practices
Chapters
Full transcript
The complete talk, organized by section.
Carmen DeArdo
This is the workshop on modern technical practices. Thanks, everyone, for attending. Terri Potts was a member of the team. Are there any other team members that are here?
Okay. One of the things, I've facilitated a few of these sessions already, and I think some of the discussion in general is around, well, where do you start, right? And we know that it's more than just the practices, right? There's the culture and everything else, the organizational change that you heard about from Courtney and her team, and other things. Even your organizational structure, that was another focus.
But where do you go? So we laid out some thoughts of the technical practices and some cultural norms, and then the paper we produced has five case studies that just show different journeys that these case studies went through. And what you'll see is, you'll see some similarities, but then there's a lot of differences, just because there isn't really one path of how to go about this, right?
I think in general, if there's any lesson, it's around find a way, sort of have a true north, find a way to get some traction, get some successes, build on those successes, and try to take advantage of the currents that are with you, right? Because we all know there's going to be a lot of currents that are coming at us and things.
So from a technical practices perspective, these are some of the things that we talked about, right? So probably not a surprise: automation. Automate everything that you can, that makes sense. Some kind of testing, obviously testing, automated testing. Continuous integration, continuous delivery. Versioning, which is just not your code. It's your test scripts, it's infrastructure, it's everything that you can to being under version control.
Visibility, some of the things Courtney talked about in her talk this morning around visibility of the work. It's important in Agile, it's important in Lean. It's important as you're doing your work here from a DevOps perspective.
Performance monitoring. How to get security and audit. Topo talked about some of those things today. The documentation, and then how are you measuring things? What's your metrics tool feedback? And then the ITIL, ITSM kind of things: incident management, change processes, things like that.
Terri Potts
Carmen, I think one thing that doesn't really jump out here that's kind of subtle but is important, and we talked a lot about it when we were developing the paper, is the architectural changes that you have to do in order to do the automated testing if you're coming from a legacy environment. Or the architectural changes that you have to make for APIs.
I know my organization, perhaps yours, is coming from big monolithic systems and trying to now move to the cloud. So that's one thing that we really didn't do a great job highlighting so much here, but don't forget about that.
Carmen DeArdo
Right. And I actually think in the document that's being produced, there is a set of those architectural practices. So the things that we've heard about as far as APIs and how do you get that monolithic application into a state where you can make small changes more quickly. So that's a great call-out, Terri.
And then some of the cultural norms about having teams that are really self-empowered, trust, trusting your teams, trusting the results of your tools. Sometimes we do automation and testing, and it's like, "I don't really know if I trust those results. I think we've got to do some more stuff over here."
The whole idea of running experiments and risk-taking, collaboration, continuous learning, one of Gene's passions, obviously. And then again, all these Lean concepts because, really, sometimes I think of DevOps as trying to realize the promise of Agile. Agile talked about going through iterations and having production-ready code. But we know that you have to do more than what's typically an Agile transformation to actually get that entire value stream running in a way that you can automate that and have visibility to it.
So here's just a couple examples of case studies. One of these is yours, right?
Terri Potts
Yeah, but we don't have to--
Carmen DeArdo
Oh, that's right. Forget what I just said. Strike that from the... Stop the tape.
So in one case, it started with APIs, and then went to a full Agile stack. Some of these you might think, well, gee. Then they really went more into automation after that. So they were doing some of the management and some of the practices around Agile and then went deeper into it, and then realized, okay, we really have to Lean out more than just Agile in order to make this work.
Government agency, I'd say I see this pattern a lot, where you're doing CI, test-driven development, blows up into more of an Agile automation, and then you build your pipeline.
Similar, I think, on case study three. There's a Lean part there in the middle that sort of is in between that, and then going to APIs and small batches.
And then the other two, again, full team stack APIs, small batches. So they were focused more on the interface and the batch size and the metrics, and then they went deeper into the CI/CD and automated test practices. And then again, I think Agile automation, more CI metrics, full stack. Sure.
Terri Potts
So one of the things, again, that doesn't jump out here that we talk about in the paper is these all look pretty happy path, right? But they're not. We talked a lot in the discussion and development of the paper, and you'll see it in the paper, that sometimes it wasn't always happy, right?
Sometimes we had some starts, and then there were some failures, and we kind of had to take a couple steps back and then take a couple steps forward, right? And a couple steps back, couple steps forward.
So I heard it after the last workshop this morning that somebody said, "Wow, I thought we were the only organization that was really struggling with this," but it was really great to sit around the table and hear that there are other people struggling with the same things.
So the important thing is that you just keep learning, and you keep taking those lessons back to your next instantiation, and you just get better and better at it.
Carmen DeArdo
Yeah, and I think that's a great point, Terri. It's a journey, right? And there's no right or wrong. We're trying to sort of give some views of various ways to get there, but it's going to be different, right?
Some areas already may have Agile, and other areas don't. Some may have a good Lean concept that they can build on, right? Some may already have good CI/CD kind of technologies, but they're missing other pieces.
So I think the great part about what you're going to do here as soon as I shut up is it's a journey. Everybody has something to contribute. Nobody has all the answers, and it's a great opportunity to share those insights.
So with that, thank you very much.
Terri Potts
Thank you, Carmen. Thank you, Terri.