Live from DevOps World | Jenkins World 2019 with GigaOm, Hurwitz & Associates and EMA

Sam Fell: Hi, everybody. Welcome to a special episode of DevOps Radio. I'm Sam Fell, and I'm being joined by three industry analysts who are joining us here at DevOps World | Jenkins World 2019. Let me introduce them, and we'll have a conversation about a couple of different things. Let’s start on the end.

Dan Kirsch: Alright, so, I'm Dan Kirsch, Managing Director of Hurwitz & Associates. We focus a lot on cloud, hybrid cloud, multi-cloud, and then a lot with DevOps, DevSecOps, and sort of how developers are using multi-cloud to help put together the next generation of applications.

Sam: Welcome. Thank you, Dan. Jon?

Jon Collins: Hi, Jon Collins at Gigaom. Let’s say that I cover DevOps, it’s like the most straightforward—

Sam: Nebulous thing.

Jon: - broad, nebulous thing that you can possibly say—but I cover DevOps.

Sam: Wonderful! Alright, well, thank you for joining us. Torsten?

Torsten Volk: Torsten Volk, research director at Enterprise Management Associates. I cover the infrastructure aspect—storage, network, compute, operating system—that enables DevOps, that enables fast releases of applications, that enables, in the highest level ultimately, digital transformation, so.

Sam: Excellent. And so, you bring up an interesting point, right? Digital transformation is something that lots of people are talking about. How do you allow that innovation that companies are creating? How do you give them a way to get that digital transformation and get that innovation out into the market? A lot of companies are trying to do that by bringing this digital transformation in-house.

Are you guys seeing that as part of the work that you're doing with your clients? Torsten, we'll start with you.

Torsten: Yeah. Digital transformation is an interesting construct, right? It’s almost a little bit like artificial intelligence where nobody exactly knows what it is, but we all know that we have to, he just said open the doors of our hotel with our cell phones, right?

Sam: Right, right.

Torsten: We have to change the ways we do business, banks have to differentiate by offering more convenience—basically, every business has to reinvent themselves. And to do that, you can’t just stay where you were in terms of releasing software periodically, you have to release feature software all the time.

And that’s what almost everybody is struggling with, and that’s what leads us to this whole hype around Kubernetes, around, you know, the container megatrends that should solve all of our issues when, in reality, it’s, to a big degree, a process issue and a cultural issue in terms of spreading that news of being able to release new software and digital offerings, not just within the world of DevOps but across the whole organization.

Sam: Right, and getting that, when you're doing those releases, making sure the rest of the organization is empowered and ready to receive it is also an important part of that whole digital transformation. If you're doing innovation faster, I think, Jon, you wanted to say something about innovation.

Jon: Yeah. I mean, we were talking earlier about digital transformation and I think what we were just saying, we don’t necessarily like the term digital transformation because it implies you're gonna end up in this wonderful place where everything just works and it’s all digital. It’s nebulous, it’s fantastic.

Sam: Are we there yet? Right, yeah.

Jon: Exactly. But maybe a better way of looking at it is, it’s about stopping doing the stupid. It’s about—it’s about taking away the things that are slowing down. For example, connecting better with your customers, working out what it is you're trying to innovate with, and delivering that much faster than you were before. Engaging the business—the business is already using technology. The whole shadow IT thing is largely because the business is wanting to get on and deliver things.

Sam: Go faster, right.

Jon: But that’s causing other problems like a thousand pipelines for a thousand applications. So, a lot of it is just stopping those problems actually getting in the way and slowing you down as you continue to try to innovate.

Sam: Find the sand in the ointment and just take it out before you spread the salve on the digital transformation motion.

Jon: Basically. I mean, we've created a lot of problems by trying to do things faster, and now, what we should really do is stop and think and work out how to do it fast and to do it right.

Sam: Nice. Okay. Dan, what are your thoughts?

Dan: Yeah. Well, it sounds like we're all talking that there needs to be some cultural change. So, really empowering the technology leaders, but also business users and business leaders also need to be on the same page. Because you can’t—you know, if business leadership says, “We need to be digitally transformed tomorrow,” you know, the IT team is gonna say, “We're 10 years behind, so this sounds great.”

Sam: [Laughter]

Dan: But we need the tools, we need the infrastructure. We need to use multiple clouds, and we also need to understand our data. So, we have different business units, they all hold onto their data. We have to break down these silos, but in a governed, controlled way with guardrails so that we have freedom, but we don’t go off the deep end.

Sam: Right. You have to empower the people at the edge to be able to innovate more, to make choices that may go against the grain of what you've already done, right, because you're experimenting. And hopefully, you fail fast if you fail, but if you don’t fail, then you want to move it through the pipeline and you wanna get it out into a place where users can benefit from it in a way that is safe and is not gonna cause any trouble for you from a governance perspective or from a regulatory perspective or from whatever. Absolutely.

Torsten: Yeah. And specifically, data, that unavailability of data suppresses, in my estimate, about 80, 90 percent of artificial intelligence projects where people say, “I can’t really get the data, and if I can’t get it, I can’t get it all, and I don't know if I can get to a proof of concept with what I have.” But to give me the data, those guys want a proof of concept, so catch-22, and—yeah, it’s not available to my DevOps process, right? I can’t put those beautiful models out there as artifacts and use them across the organization.

Sam: Right. Jon, did you have something you wanted to add?

Jon: Yeah, that’s the truth. I mean, I wonder if we're at a stage where, a bit like back in the day when suddenly patterns were a thing, so there were all the design patterns and analysis patterns and so on—and process patterns, for that matter. What we need is, rather than having a thousand different ways—when you're in an experimental place, everyone’s trying to innovate, everyone’s got the same tools, and everyone can deliver something very quickly. 

Sam: Right.

Jon: But then they're all hitting the same problems, and these are all solved problems, in a lot of ways. So, even if not inside your own organization, more broadly, someone will have come up against that problem and solved it.

Sam: And solved it.

Jon: So, essentially, what we need right now is, I would say, a bit more than guardrails. We need something that actually starts to give you the components from a pattern sense of something a bit more specific than, “Here’s the policies of saying, ‘Work within that and you'll be okay.’” What do we then say about how to work within that so that you'll remain okay.

Sam: Well, you're making me think of Darwinism and evolution.

Jon: [Laughter]

Sam: Seriously.

Jon: Yeah, yeah, yeah—definitely.

Sam: Right? So, you've got all these teams and they all have the same basic ingredients and they're all coming up with a different way of solving the problem. Some of those ways of solving the problem are gonna work really well and some aren’t. You don’t wanna have to go back and evolve into a cul-de-sac again, right? You don’t want the business to say, “Oh, let’s try this thing that didn't work a long time ago.” You wanna say, “Here’s a pattern that works, here’s something that we have proof that works,” and then be able to, as you said, to sort of measure it and monitor it, understand the work that we just did—how did that impact the business?

Jon: Mm-hmm.

Sam: What was the downstream effect of all that work that we just did, and was it beneficial, or was it not beneficial? Yeah, maybe we went faster or maybe we went slower. Maybe we had more or less known quality in what we're shipping. Because sometimes you can go faster and have less known quality, right, as long as you're getting out there.

Jon: Well, a good example would be the metrics. Like, so—just give me the five things I should be measuring right now. Just tell me what those five things are, and -

Sam: Well, you guys are the analysts. You should tell everybody what there—

Jon: No, no, no—exactly. If there’s a sixth, then fine, I can do the sixth, as well. But let’s all just agree to those five. That would be great.

Sam: That would be so much easier, then, if every business was exactly the same.

Dan: Well, I mean, I think the push to microservices and modularity is exactly—you know, you don’t want to reinvent the same access and identity application 25 times for every application. You have something that works, just repeat it across the business. And if that’s broken, you can fix it, and then it just gets pushed through on every app.

So, I think businesses are starting to realize that we don’t need every team to reinvent the same thing. So, I think that’s really the whole push behind sort of this new way of development.

Torsten: That’s a very big part of a developer’s day is to do the boring stuff, right? To do the authentication authorization and all the basic things, and then—

Sam: To get to the actual business value, right.

Torsten: - right, and business value takes only 20, 30 percent because then supporting the infrastructure takes another big part, and that’s where the patterns come in, as well.

Sam: Yeah. So, Dan, you asked an interesting question that I think we'll end on. I don't know that we have a ton of time left, but you mentioned there are things that you think are—we have a little bit more time. We've got 10 more minutes. Look at that—wow! You guys are talking super fast. [Laughter] That’s wonderful.

Jon: There are 17 things you need to know—and that’s them.

Dan: And that’s it, yeah. Charge by the word.

Sam: If you stick around to the end of the episode, you'll hear what those six metrics are that you need to know, they're gonna tell you. Just hold on one second.

So, what are some of the things that you're seeing in the ecosystem that’s getting a lot of attention that you think doesn’t necessarily deserve all the attention? Where’s the hype in the market? 

Dan: So, what I was thinking is that there is—

Jon: There is a lot of hype in the market.

Dan: - yeah, there is a lot of hype. What we've seen is, there’s just so much about automating poor processes. That’s sometimes a nice first step in a digital transformation, but oftentimes, it’s really a Trojan horse—they get a consulting firm in there, they'll put RPA over some processes, you might have a process that’s 20 years old and you're going from screen to screen to screen and that’s great, you put RPA on, it’s gonna help your customer service reps not ask the same questions over and over again. 

The reality is, your system’s broken—and yeah, you can put another Band-Aid on it, but you're making your infrastructure heavier, less modular, and all of a sudden, it’s even more parts that break and it makes it harder to break down.

Sam: More brittle.

Dan: Yeah.

Sam: Yeah, yeah. Interesting.

Dan: And expensive.

Sam: And expensive, yeah, and there’s a chance for redundant information that’s not being updated correctly. There’s a whole host of problems that you could introduce when you're not leaning out the entire value chain, and instead, you're sort of trying to put a Band-Aid on one part of that overall holistic system. Yeah, interesting. What about you? What do you find hyped?

Jon: I think, slightly different to hype, but kind of the same—something that’s kind of outstayed its welcome a little bit is the whole notion of continuousness. So, we've had this idea that “Oh, how many deliveries do you do a day?” “Oh, I do 10.” “Well, I do 20.” “Oh, I do 50 deliveries a day.” “Fifty? That’s nothing. I've got 500 deliveries.”

What are they? Nothing. Who cares? You know, a lot of it’s rubbish or it was nothing to do with the customer demand.

Sam: No one actually noticed, right.

Jon: No one actually noticed. Was it secure, was it—heck, because you can’t without automation, which we talked about before.

So, the idea that continuousness for its own sake—easy for you to say—has outstayed its welcome, I think. To actually have the ability to repeatedly iterate and develop and innovate and deliver additional value by bringing the right people in right up front, but also by putting the right policy criteria around it, that should be the way that we think about it moving forward, rather than just continuousness.

Sam: So, I'm gonna challenge you a little bit on that. I'm gonna say, so, the idea of shipping all the time—so, I wanna make sure I understand—so, the idea of shipping all the time is not necessarily the most valuable thing. The idea of shipping value every time is what we should be focusing—

Jon: Which comes back to things like value stream management, which is just applying the value metric—there’s one of your metrics.

Sam: Right.

Jon: So, two metrics—efficiency of, are you actually doing things without additional costs that you shouldn’t be spending? And then effectiveness, business value—is it actually giving you better customer relationships, are you selling more products, are your internal and external stakeholders happier as a result?

Sam: Are you doing things right, and are you doing the right things?

Jon: Correct. Those two.

Sam: Yeah. That’s very important.

Jon: It’s good stuff.

Sam: Very important things to focus on, right. Torsten, what are you seeing that you think is maybe a little bit too hyped right now?

Torsten: So, this needs a disclaimer, because I'm saying Kubernetes. [Laughter] 

Sam: This is gonna get some views. This is gonna get some trolls at the bottom.

Torsten: No, no—I don’t hate Kubernetes, right?

Sam: [Laughter] 

Torsten: It’s 95 percent of my time that I spend every day is looking into all the different aspects of Kubernetes and answering questions about, you know, all the nitty-gritty details of customers trying to turn Kubernetes into another hypervisor, instead of using it the way it was intended to be used, right? Transformation to microservices and just getting rid of the old stuff instead of trying to look how you can use the existing infrastructure and put it into containers first and—you know? It’s almost, you look at all of those Kubernetes conferences, it’s really, you have to almost be there as an IT guy, right, right now. Because that is the next big thing.

But in reality, there is—as we said, you know, automation and application patterns and maybe DevOps process patterns are something that are much more important because Kubernetes, if you use it just like another hypervisor, that’s—

Sam: Another run time, yeah.

Torsten: - you know? We are going 10 years back because that’s about the feature set that it has compared to vSphere now, so.

Sam: Right, exactly. And you think about where—I was talking with somebody earlier today about sort of VMware and virtualization and IT and executives were very interested in virtualization, and those things 15 years ago, 20 years ago, maybe? Is that when that—

Torsten: Twenty, yeah—’99.

Sam: Because that changed the game for how you operated your data center, and we're seeing that same—now, I mean, virtualization is still incredible, but we're seeing that same excitement around Kubernetes, and you're saying that there’s too much excitement about Kubernetes because it’s not as different an animal as people think it is?

Torsten: I would say there’s too much hype around Kubernetes, because all that containers do is, they let multiple applications securely share the same operating system, the same Linux operating system, right? But then when people don’t actually use namespaces inside of Kubernetes and they say, “Ooh, let’s build another cluster and another cluster and another cluster”—at the end of the day, I have the same that I had when I had a hypervisor—

Sam: Exactly.

Torsten: - but an additional layer of abstraction and management tools.

Sam: Right.

Torsten: That’s what I don’t like. Interesting.

Jon: It’s interesting, because it kind of misses—to me, the great thing about microservices as a concept is the fact you can now architect scalable, massively distributed applications, which was very hard to do before, and now you've got the opportunity to do it. But it only gives you the opportunity. You can still write—it’s like real programmers can write Fortran in any language.

Torsten: Yeah.

Jon: You know, you can still write—

Torsten: Spaghetti code.

Jon: - spaghetti code, you can write a mainframe and stick it in a container and jobs are good. But it does give us that opportunity, and it’s a shame that there’s so much hype around the thing, Kubernetes, and it’s kinda distracting from, we've got an opportunity here to build well-architected, distributed, scalable.

Sam: Right, and thinking about it as an abstraction or as an architectural element as opposed to just—

Jon: Yeah. I'm back on patterns again, sorry.

Dan: Well, and that’s another thing—I don’t think Kubernetes was meant to be a massive platform for, you know, the mass majority of developers to be using. I mean, it’s complicated, it has a high barrier of entry. I mean, we're at a Kubernetes conference, so people are very interested in it, but the reality is, not every developer needs to know the ins and outs of Kubernetes. There’s gonna be—there already is abstraction built on top of it, but you know, the next three, four, five years, there’s gonna be more layers of abstraction, automation, there’s gonna be more and more automation.

So, it’s not—people aren’t gonna really need to know all about Kubernetes. You know, they'll wanna know that Kubernetes is sort of an underpinning platform on the platform that they're developing on.

Sam: Right. Well, I hear our—we got a keynote coming up, and we're actually, it’s the CDF keynote, which is pretty exciting, and they'll be talking, I believe, a little bit about Kubernetes, which is very exciting, and it is something that a lot of people are using.

To your point, developers may not need to know how to use it. Operations people, one of the things that they are having trouble with is understanding this torrent of new technology that’s coming off the Dev side, all this innovation—to your point, all this experimentation. Will this work, will this not work? I have no idea. If it does work—guess what, Ops people need to support it. And so, finding ways to help these folks understand this new technology and take better advantage of it without having the risk of having, you know, systems in their production environment that they don’t necessarily understand. That’s not a great thing, right? [Laughter]

Jon: You mean, more of them.

Sam: More things that they don’t understand—exactly. Well, gentlemen, I really appreciate you taking the time to sit down with us and talk a little bit today about digital transformation and innovation and Kubernetes and things that are on the hype cycle. Thank you, all, for watching another episode of DevOps Radio.

Like what you’ve heard today? Don’t miss out on our next episode. Subscribe to DevOps Radio on iTunes or visit our website at CloudBees.com. For more updates on DevOps Radio and industry buzz follow CloudBees on Twitter, Facebook, and LinkedIn.