The convergence between Software Development & Operations has had a deep impact on the technical stack used by startups & industry giants alike, with companies like Chef gaining prominence since the term DevOps came into parlance almost a decade ago. We talk with their CTO Adam Jacob about the meanint of the trend and why, like many successful technical waves, it’s been shaped by business needs.
Tim Anglade, Executive in Residence at Scale Venture Partners: Cause I feel like I started hearing the term a long time ago, but it’s never kind of quite clear.
Adam Jacob, CTO at Chef: Yeah, like when was the exact year? I don’t know, like 2008 maybe.
Tim: Yeah maybe.
Adam: 2009?
Tim: I remember being on Offcon and like Chef being one of those companies that was like out there
Adam: Yeah
Tim: The word kind of coming out of there.
Adam: There were sort of two seminal things. So Patrick Debois, who’s a dude in Belgium wrote a blog post about what he was seeing in terms of this rise in cooperation between developers and operations people. And then John Holspot and Paul Hammond gave a talk at Velocity that was called like ten deploys a day at Flickr a Dev and Ops cooperation story or whatever. Those two things together are what got you to the word, like calling it DevOps, and I think it was Patrick who first said, first put the parts together and sort of made that what it was. That feels like it was roughly 2009. What’s interesting about it is I think you have lots of different people. They didn’t know each other, Patrick and John, and certainly the guys at Google who had been developing the SRE program, they didn’t know them either. And there’s this, it sort of happened organically because it was the same challenge everybody was facing everywhere. And had done multiple times in their career.
Tim: Right and it’s interesting cause I remember around the same time there was also like the no sequel turn and the sequel wave. And it was similarly kind of ambiguous read, like what does that even mean, and is that the right term. And I remember back in those days thinking that DevOps was such a bad name.
Adam: I think it was a terrible name. Yeah it’s and awful name.
Tim: But what was good about it I think is that it captures the essence of developers and operations and how they were getting closer together. And it wasn’t very descriptive as to what the technology underneath it would be but it turns out that the technologies underneath it changed quite a bit since the term appeared.
Adam: I think the term’s fine as a thing that describes cooperation that we require. What sucks is that what if you’re a security guy and you’re like, well I’m not a developer or operations, so now do we have a DecSecOps now? We have to collaborate too. So that was weird. I don’t know what a better word would have been that would have been more helpful. I think the fact that it’s fundamentally a cultural thing first, so there is a tooling change and that’s very dramatic but that tooling change is in support of the cultural transition right? So there’s this culture you have to change. If you have tooling already, and what company doesn’t sort of have tooling that supports their workflow, and if you want to change how people work you have to change the tools and so they go together. So I don’t know how descriptive it is in terms of generation of tooling. But in a way it’s descriptive enough because you can put anything that encourages that collaboration and helps bring those people together into a single work flow. Into a single set of behavior. Those are sort of DevOps tools.
Tim: Right, and I’m very very kind of intrigued by that because of course a lot of stuff happened, like cloud happened and a lot of different things that you could say are their own distinct kind of machines and waves but it does seem DevOps, as a cultural saying has had a very far reaching impact and I would like it personally, and maybe you would agree, to stuff like micro services function of the service like this idea of like this line between writing code and running code and all the people that have to do a mix of that. How much that has changed. And if you look at a startup, even if you look at a big company in 2008 versus today.
Adam: Yeah, it’s changing dramatically. I think it’s more interesting to look at big companies than startups because startups are small and you can always be wacky. You find a venture investor who’s willing to let you be a little nutty if the returns are good and who is not? But revenue covers a lot of sins there in venture land. But if you look at large companies, you absolutely see these really dramatic changes in, you look at companies like Alaska Airlines who talks about being a 60 second airline. Where everything that a customer goes through needs to complete in 60 seconds. And if you ask them what’s integral to making that happen, they’ll tell you DevOps is one of those most integral things like we can’t do it without this super high degree of collaboration and coordination and software and operations and it’s really fundamental to what they’re doing and you see the same thing at Nike, and at giant banks and it’s super.
Tim: It worked really on two levels, it changed the internal engineering culture now that you have in companies, but it also really facilitated, in my opinion, this transition where now a lot of industry that weren’t about technology in a strict sense, or that kind of technology, not about software become increasingly reliant on software. And you said airlines which is a big example. Wasn’t it Delta or some other airline that had this gigantic screw-up right? With their automation and everything else. And that can really kind of nail you business. It’s not just the Amazon and the pure software companies of this world.
Adam: No its every singe organization. In a way that I think is super different. And when you look at DevOps coming out of the people who really felt it first, we were all web innovators. The one thing that’s true about all web innovators are they are exclusively software companies. Our businesses only exist as software. What’s becoming increasingly true is that everybody’s business basically only exist as software. Especially when you look at something like consumer banking. If your consumer bank has a poor online web portal, and you can’t do online banking, you will switch banks. Because you don’t want to go to the bank. They’re literally using it as a competitive weapon.
Tim: It’s a bit of a chicken and egg thing right? To dome extent DevOps helped companies make that transition but there’s also this incredible market demand to go to towards fast integration software even if you’re an airline, even if you’re a bank.
Adam: I would argue that the macro trend, the market pulled us, we didn’t push it. That the market was pulling those companies in that direction, and then when you stand up and look around you work in a mutual company, you work in a bank, whatever and you’re trying to figure out what should we do it’s natural to look at prior art. Well who’s good at this? Who goes faster, what is competitive in this way. And what you find when you do that is big web companies. And so it’s sort of natural that that happens. What’s interesting there is that there’s an argument that’s happening in industry right now that I don’t think is particularly true which is that therefore what is going to happen is that all those big companies will wind up looking like Google or they’ll look like Facebook. And I think that that is wrong. Partly because it sells short the difficulties inherent in running those industries. It’s not the same to run Morgan Stanley as it is to run Google, it’s just not. And that doesn’t mean that you won’t have very similar principles, very similar beliefs, very similar technology even, but will the technology solutions be identical to Google? How long will it take before we rebuild something like SAP core banking? In the same way Google built it. Is there even value in building it that way? Does it ever get big enough that you care? And we’re not asking those questions yet. We’re still asking the first question which is holy crap how do we go fast enough to be competitive. Were not actually asking the question of how am I going to build that technology in a way that suits the business or is good for everybody?
Tim: That’s true and that’s a big opportunity for entrepreneurs because there’s kind of two tracks happening. It’s happening quite a bit. But in DevOps it’s particularly visible where you can’t have the Google, Facebook your startup led type of open source projects. And then you have enterprise players, enterprise startups. That are able to factor either those facts or those concepts and sell them to this kind of more stodgy, established audience. And you see a lot of companies like that for Kimani’s project, or Singa that catches up. You can usually find five to 10 enterprise types that are doing very very well. Kind of selling a variant of that to the enterprise. I think that’s something to remember if you’re doing a startup, it’s not necessarily about the pure play cool open source software stuff you could be doing, although it certainly can be and it is successful. There’s also kind of different ways of looking at a problem and matching it to the audience is usually a big problem.
Adam: Yeah I think in terms of entrepreneurial opportunity and venture opportunity I think that there’s an incredible opportunity to really think differently about how we design a product to the needs of those organizations. In the same way when you look at the one thing that’s true about Google and Facebook and Yahoo and all those companies, even though they have dramatically different cultures, and they have very different technology stacks. One thing that’s true, they all maniacally focused on their own workflow, on their own needs. It’s the way Google works now. And asking the question well, okay how will these enterprises work now? Asking that question requires you to actually understand how they work. And when you look at how developers talk and how, especially in the big lab, we tend to talk about those companies as legacy, we talk about the software that they run like legacy. What we mean by that is all the software that runs all the things in the entire world. So the entire world economy runs on legacy software. And we talk about it like it’s this rotting pile of garbage. Its just fascinating, like maybe some of it is, but like really? There all just the dumbest people who lived on the planet? It’s a very poor product position to start out from to say if only all these people stopped being stupid, like how they’ve been for however long they’ve been, and just do this thing that is completely different then everything would be cool. Maybe that’s the only way, and maybe they are. But it feels really unlikely.
Tim: It is such a big thing to think about not only solving a technical challenge but solving that business fit. How do you get people, we’re in this situation, we have the kind of stack, we have this those business concerns, and get them to adopt this. And I think if you look at software in general you can find a lot of companies, like Toyo for example or PubNub or even Amazon right? Where a lot of the innovation wasn’t so much technical they were about the business way.
Adam: They were about delivery. It was about the user experience.
Tim: Instead of buying servers you can rent them. Sort of like hiring consulting to do this you can buy this-
Adam: And it was definitely technical innovation back there, but that’s not what really made it. Especially EC2, like renting a server is an interesting one. Our first pass as trying to do that for large enterprises was like Open Stack. It was the private cloud. Those have largely failed. If we really go look at how many of those deployments are really successful, and how many people are really happy with it, its not nearly as much as you would have hoped. And so the idea is that that was going to become the basic primitive for on premise gear inside these huge enterprises to look right now that that’s not going to be true. So what’s right the design? Because it clearly isn’t copying whatever AWS did because it’s not quite the same right? So what is the behavior that you want and what is the interface that those people want and we don’t know and that’s where the real entrepreneurial opportunity lives.
Tim: And conversely you have the other way around where largely developers have moved on from past in a large sense and Heroku kind of way of doing it, but now that’s super popular in the enterprise and kind of having a second life.
Adam: You can see the lag. I’m sort of a vocal PaaS doubter. I think in general I tend to believe more in human beings and what they feel and think. And a good example is, people don’t like to be put in a box. And they definitely don’t stay there when you put them in. And so I’m buying a house. And it’s a beautiful house and I’m really excited about it. But what it isn’t is a manufactured home. You could go and buy a plot of land and stick manufactured home on it and the catalogs are huge and they have every possible feature you could ever want in a home. And they’ll just snap it together, it will be done in like a month, and it will be amazing, but no one will do it. You don’t do it. You don’t do it because you want craftsmanship. You want to believe that there’s a human piece there. And when you see that in operations and in business, that’s what PaaS is telling you. They’re like hey, if you rewrite the world to match this platform everything will be good for you. And it’s like okay, but for how long are you gonna stick there?
Tim: No absolutely, I think you have to find your section of having that technology, fitting it to enterprise, and keeping that creativity and spirit in the middle and I think that’s so important.
Adam: You’ve gotta have it and if you don’t you’ll fail. Or for a minute you’ll feel successful, because you’ll be floating to equilibrium, and then you won’t win because everybody forgets that these huge businesses compete with each other. They’re trying to win. And so the idea that they’re gonna standardize on the same platform and then they won’t innovate at all, on technology ever again, like come on that’s not true. One of them will decide to innovate, and they will therefore win. Which means all of them need to have the ability to innovate.
Tim: That makes sense. Thanks a bunch.
Adam: Yeah no problem, thanks for having me.
