Skip to content

The problem with dev/ops culture – IT Management & Cloud Podcast #083

John Willis

Recently, people have been talking about dev/ops as more than technology, but as culture. In this episode, I play a bit of the irate straw-man to suss out exactly what that means in more detail. We recorded this during SXSW 2011 at the lovely Driskill bar, so pardon the background noise.

Download the episode directly right here, subscribe to the feed in iTunes or other podcatcher to have episodes downloaded automatically, or just click play below to listen to it right here:


As usual with these un-sponsored episodes, I haven’t spent time to clean up the transcript. If you see us saying something crazy, check the original audio first. There are time-codes where there were transcription problems.

Michael Coté: Hello everybody! It’s a special edition that we are recording of the IT Management Cloud Podcast. We are here at South by Southwest in the lovely Driskill Bar, where dedicated podcast listeners will know we recorded, not you and me necessarily, but I have recorded many episodes in here. So what do you think of the South by Southwest so far, John?

John Willis: Pretty cool, makes me want to live in Austin.

Michael Coté: Well, you actually said you wanted to live in like San Marcos or something, right?

John Willis: Yeah, so it’s —

Michael Coté: Because you are staying down there.

John Willis: Yeah, I am staying there, because I couldn’t get a hotel last minute, so I am staying at a hotel in San Marcos. I actually flew into San Antonio because the price of airfare was half price. On the way there I was thinking, being halfway between San Antonio and Austin might not be a bad gig.

Michael Coté: Yeah, definitely! Like I was saying, there are a lot of people who live up and down I35 there.

So you got here yesterday and ostensibly you are some sort of like IT cloud guy, and as I recall, the South by Southwest is about like Foursquare and Twitter. So how are these things joining together for you? What’s going on here?

John Willis: Oh, jeez, hit me out with a question. I don’t know for any of that stuff, Michael, and by the way, this is John Willis.

Michael Coté: That’s right.

John Willis:

Michael Coté: You have probably been drinking plenty of the water, picking up the —

John Willis: They got a Bloody Mary bar here and it’s only 1 o’clock. So we are doing the Cloudies Award Tuesday night. So I primarily came in for that.

So I had, near the end of last year, Dave Nielsen, who runs CloudCamp had asked me if I would want to put the Cloudies, kind of contribute it — I don’t know if people know me, I run these really just silly Cloudy Awards in 2008-2009. I give out awards for somewhat meaningful, but somewhat silly things, and David asked if we could do it through the CloudCamp.

So he is having a big old party Tuesday night by the Dining Guys (ph), they are sponsoring it. So I am coming in for that, to kind of be the co-host of that. But I looked at the schedule and there was some really cool kind of party events.

Michael Coté: Yeah, like last night we went to the Etsy Ops con fab. I have been trying to use the word con fab a lot. I don’t know what that means.

John Willis: So Etsy was — they did a presentation on how they do continuous delivery and code by design I guess is what they call it.

Michael Coté: It was pretty good. It was detailed with — it could have been a lot more detailed, but it was detailed enough. Then they spoke for like an hour.

John Willis: Yeah, it was enough to cover in an hour. Probably if you have never really kind of seen the gory details of all that stuff, it would seem kind of confusing at a high level. But they went through — Etsy’s John Allspaw, the kind of king of DevOps most people consider, was the guy at Flickr who kind of started, not started, but really drove the multiple deployments today, at Flickr, and really kind of changing the way people look at just deployments in general.

And he has taken a part of the team that was at Flickr and himself, they have really done it to perfection at Etsy. They went over this, kind of idea, they do like 30 deploys a day.

Michael Coté: At Etsy?

John Willis: Yeah. That continuous delivery model.

Michael Coté: Did they say what’s in those deploys, because I mean, I am not an Etsy user, but I don’t feel like I have heard that there are like 30 new big features that they have?

John Willis: No, and that’s the whole thing. That’s the whole idea of this continuous delivery is small chunks of code. A lot of — there is a lot of philosophy behind the continuous delivery model, things — when people see it first they are like, I can never do that, or my God.

When I present — I have been doing some presentation on continuous delivery, and the first thing I do is I ask people, how many people have seen the movie ‘Saving Private Ryan’? People raise their hands, and I am like, remember that first five minutes where you just thought this was the most horrible thing and you were going to walk out of the movie. That’s what I tell them, like be prepared for this when you hear about doing 50 deploys a day.

But yeah, I mean, they do small coaching. It’s an idea of getting completely out of this idea of like building up releases, even like two week releases, and always delivering from trunk and I think like 15 lines of code would be a production deploy.

Michael Coté: I guess, I came in maybe a third of the way through when they were talking about dashboards and monitors a lot.

John Willis: You were eating a burger at that burger joint?

Michael Coté: That’s right, Casino El Camino, and I had to wait an hour for that. It’s a good burger. And I should have gotten a medium rare. It’s a tip, don’t get it medium, get it medium rare, because it’s juicy anyways.

The part of the safety net they went over was as, they have a lot of monitoring and log analysis and stuff going on for all of that stuff. So they are basically deploying knowing that they are going to need to be checking for errors and finding things like that.

The other thing that was interesting about it, and John is probably chuckling to himself a little bit, because after the talk I railed against how it was a terrible talk. But that’s about it.

John Willis: You didn’t say that. You had a lot of questions about it.

Michael Coté: That’s right. But another thing that is clear is that, they use a lot of sort of custom things, because they are using a lot of advanced — well, advanced technology is the wrong word. There needs to be a word for like using MongoDB or using a NoSQL technology or using — and it’s not unsupported. I don’t know what this body of stuff is, but it’s sort of like open source infrastructure that’s not mainstream, right?

John Willis: Yeah.

Michael Coté: They are not using like WebSphere or they are not even using Tomcat for things that vendors would be supporting so they have to fit out a lot of that stuff to monitor on their own. And consequently, they have built, as you were saying, some nice monitoring things like Statsy and other stuff.

John Willis: Well, what’s interesting; a couple of things. The one point I wanted to make about their ability to do delivery. I mean, I think any company that you see that talks about doing this continuous delivery, the first thing they are going to talk about is, there are some really important points like, don’t even try this unless you are going to invest heavily in test driven development, and you are going to really have tight code coverage.

And then some companies even try to start thinking about immune system, like predeploy immune systems and really going all out on kind of behavior driven monitoring, behavior driven or test driven development and all that. So if you have that stuff, then this idea of being able to do that, then you put yourself in a kind of a bulletproof position.

But to your point about just technology in general, it’s funny, I think that it’s — we are seeing kind of — it’s funny, when we first started talking about like open source tools, there was this kind of migration of open source. Early on, we saw things like MySQL and JBoss and stuff like that. And everybody was like predicting the next wave to be the IT infrastructure stuff, and in some ways it was kind of a false start.

We saw Zenoss, and not that those aren’t great products, but we saw this wave, but we didn’t solve anything. What we are seeing now though is this like third wave of not so much product with technology, it’s technology that’s leaking out from large corporations.

Like, for example, one of the things that made Statsy interesting is a time series database called Graphite, that was developed at —

Michael Coté: Yeah, and you have nothing but praise for that; I need to look into it.

John Willis: It’s pretty cool! It was developed at Orbitz. So what we are seeing is this leakage of things that large — in this case it was Orbitz, but large companies like Amazon and Google and all these companies that kind of did their own, they wrote their own solutions from massively scale infrastructure and we are starting to see a lot of variations or leakage of those technologies that they had to build, because there was nothing out there to solve their problem.

Now it’s becoming commoditized. I mean, this is a — to take a program Chef, that’s what we say is, not that Chef came from Amazon, but a lot of our guys came from Amazon. So we like to think that the kinds of things that Amazon were doing, we have commoditized.

Michael Coté: Yeah, and building on that. We have talked about technology from high scale web stuff for a while now, and I think what’s interesting is there are companies trying to figure out how they would apply that to the way they run their business.

For me like that’s kind of an interesting thing to sort out, because it is interesting to think how — wow, that’s a tall Bloody Mary.

John Willis: Yeah, they do a good job here.

Michael Coté: They left about a fourth of a cup space for you to put some tomato juice in.

John Willis: Well, what’s cool is they have got like — I guess because South by Southwest is international, they have got the clamato juice. So man, you put me in heaven here.

Michael Coté: You can take a Michelada, maybe you should try that.

John Willis: Because the only place I ever get like clamato or what they call a Bloody Caesar is in Canada. So I am in heaven.

Michael Coté: Yeah, you should try, I mean, after this one, a beer and you can have a Michelada.

John Willis: Well, the thing is the bar here is, you have got to take a picture, they have got like every — you name it and you’ve got like asparagus — so that’s my lunch, man.

Michael Coté: I mean, you are pretty much set, right?

John Willis: I am just piling it on. So, yeah, just piling in all the vegetables with your Bloody Mary or your Bloody Caesar.

Michael Coté: So anyways, I interrupted myself, because I realized that talking about Bloody Mary would be much more interesting than whatever I was rambling about.

John Willis: Bloody Caesar.

Michael Coté: That’s right. But yeah, that’s the thing that I am looking forward. I just want to see how normal people use these technologies.


It’s almost like — and there is a good confluence of the demand for mobile applications to do SaaS based or cloud based things that I think kind of drives the need for that. They didn’t exist so much when these high scale technologies were out there. So if more normal companies want to have a mobile site, whether it’s for their internal employees or external employees, and if they want to have external facing things, then you sort of need to run your stuff this way.

John Willis: It’s funny, because it’s a lot about what DevOps is all about, like is the — a lot of what’s going on now is — I get a lot of questions like how do you do this?

Like you look at an Etsy and another company that I am a big fan of is Wealthfront, used to be KaChing, and you look how they are doing this continuous flurry, and it’s amazing, but it’s like purpose-driven, you know what I mean? It’s like, how do you get a large corporation to change to that and the answer is, it’s hard, because it’s cultural — it’s across the board, it’s not just the tool.

You don’t just like say, okay, you guys go use Graphite now, done, end of story, right? It’s like these guys have purpose, they are a startup, Etsy, those guys are like pretty excited about what they are doing, every one of them.

11:11, the guy who is dumping metrics into the Graphite database, he was up there. They are all pretty damn excited about what they are doing, and I think that’s the secret sauce is, how can the enterprise —

Michael Coté: I mean, would you call that sort of like a passion they have or like what or is it more that they have a more defined — like you were saying they have a purpose, like they have more defined roles of what they are doing?

John Willis: Yes. So I have been reading this guy, like I have been trying to do a little more research on kind of DevOps culture, just because I got into DevOps for the technology, pure, but what became a fascinating part of the discussion of DevOps is that I didn’t think I would get this interested and is the DevOps, the culture part of it, the behavior patterns.

So I have been doing a little research and there’s some guy — I can’t remember his name right now. He is a guy that goes into large corporations and tries to change behavior and leadership patterns, and one of things he says that he tries to —

Michael Coté: Is this one of those five effective habits or —

John Willis: It’s one of those kind of guys.

Michael Coté: One of those guys?

John Willis: Yeah. He has written like a bunch of books. I can’t think of his name right now, but he has got a blog that he runs on leadership on Forbes.

Anyway, so one of the things he talked about that I realized that — so all the stuff I have been learning about like DevOps culture and behavior, then I started reading this guy, taking it from this kind of leadership and large corporations, and the one thing that would seem common is that, you see that one of these you have got the force of this sense of urgency, and they go into big companies and tries to get this sense of urgency. It starts at the top and it kind of explodes down, and I think that’s the thing that startups have.

Startups live and die by the sense — you wake up in the morning — you don’t sleep the whole night, there is a sense of urgency. And as you grow, you start losing that sense of urgency. Like these transitions from — like you are a guy that handles like 15 different things and then you hire a guy that’s going to do marketing and now you only do ten things and then you hire a guy that does. And the sense of urgency starts becoming siloed by definition.

And that’s the beauty of these companies like Etsy and Wealthfront is, they still figure out how to grow by keeping that, whether it’s purpose or urgency, and if you don’t have that, again, like you were asking me like, well, I don’t see much in — I am paraphrasing what you were asking me about the logging and monitoring, so big deal. You collect a lot of metrics and you graph it, right?

Michael Coté: Right.

John Willis: And I was actually fumbling on how to explain, no, it really is a big deal, but I can’t tell you why. But I think it is, it has got to go back to that behavior or the culture that like, if you have got that, then the purpose of — why do you put something like, not to pick on Zenoss, but put Zenoss in and just collect data, big shit. Like, go ahead and have a purpose and then figure out from that purpose or urgency why you want to collect data, now shit happens.

Michael Coté: Yeah, yeah. No. I mean, that’s another nice thing about the Etsy presentation is, they are very application focused, which I mean, correct me if I am wrong, but my perception is, a lot of IT management is — it’s very — this is like the old saw, but it’s very siloed, in that you are not necessarily focused on an application or even a service, you are kind of focused on whatever your part is.

And to have that bigger — that sense of purpose or application, if you will, like we worry about people be on the list things to buy and sell things to buy. Or there is the old cliché about Amazon, the only thing we watch is how much people are buying, that’s our number one metric. Like that is like a nice practical way I think to connect the management and monitoring that you are doing to some sense of purpose, like why do we have all this crap.


John Willis: Right. You start from a purpose driven — like the guys at Wealthfront talk about monitoring, but they don’t think of this monitoring for monitoring’s sake, they monitor for behavior or they monitor for business behavior or business goals. They call it Business Goal Monitoring.

Again, it sounds clichéd. There is a lot to it, and I think the key — the more and more I think about it like, I think about the sins of IT software is that we just never focus. We talk the big game about process and people and all this stuff.

I get this famous saying now — famous, I made it up, so it’s famous in my head, but it’s something that you will love I know is that — people were asking me about ITIL. What is ITIL in DevOps? And I said, it’s funny that ITIL is about process over people.

Michael Coté: Right, right, right.

John Willis: Right, no doubt about it. ITIL is like, you are so broken, put this in and at least you won’t be broken. May not be perfect, but at least you are broken anymore.

But what DevOps is about — what’s the beauty of DevOps is, it’s about people first, process second. It is all about the people and then once the — even process. We don’t come in and just say, DevOps is process, rule number one, put this on.

People try to say continuous delivery is DevOps. Well, with like bad culture or like putting continuous delivery model in will be a freaking disaster.

Michael Coté: Yeah, yeah, yeah. Well, to give you a chance to eat a little bit.

John Willis: No, I am good, I am good.

Michael Coté: Let me go on a little monologue. The other thing I was railing about while we were waiting for that burger was how I always get upset about how —

John Willis: You are like — your life kind of revolves around burgers. I mean, it has taken me a couple of years to get this, but like your slide decks have burgers in them.

Michael Coté: I love a good burger.

John Willis: Your metaphors are burgers.

Michael Coté: Yeah, yeah. I think it’s just — I think Lord Sandwich was on. Delicious piece of meat and you get good pieces of bread and you basically have your side, it’s kind of a complete handheld efficient meal, and it’s delicious. And you get French fries with it.

I have a mixed relationship with French fries. I feel like French fries are kind of a crap food. Like I could be spending my time eating something better, and yet, they are delicious. I am not a big fan of the potato. That’s the whole thing.

John Willis: What’s like — yeah, I think I agree on the potato. I mean, you can have a good two pair carrot, and the hamburger is like the top dog. Maybe it’s like Batman and Robin, right?

Michael Coté: Oh yeah.

John Willis: The fries are Robin.

Michael Coté: It’s sort of like — like so for example, let me highlight why I think French fries are — I think French fries are — they take up space that could be taken up by something better. For example, have you ever had like sort of deep fried Polenta Sticks?

John Willis: No.

Michael Coté: See, that’s good. That’s a good sort of thing. I sort of like — I don’t think it would cost anymore to have Polenta instead of like French fries, if it kind of had the same supply chain mechanics behind it, right?

John Willis: Right. It would really screw to say that, but yeah, it would e all right.

Michael Coté: Anyhow, so we were waiting for the burger and I was saying how — I get upset when DevOps people are talking and things are going on, they are like, well, it’s just all about culture. Like DevOps is all a cultural thing, and of course I exaggerate to make a very narrow point. So it’s much more nuance than that.

But what I do and let me think, the reason that I like that is because back in the, I don’t know, late 90s, early 2000, when people were like arguing all hot and heavy about agile development. The same way they early on are kind of arguing about DevOps and talking about it, like the same thing would come up and it would drive me crazy.

You would be talking about like, well, how do we do this and how do we introduce it, and this, that, and the other, and people are trying to figure out like — well, a large part of people were trying to shoot down the idea because they just want to keep doing what they are doing.

And other people are like, I really like these ideas and I am going to go back to work, and it’s just not going to work, and like help me figure out how to make it work.

And then the thing that would — kind of like a dead end of helpfulness would be like, well, it’s really a cultural issue for agile. You have got to change the culture. You have got to — and then as an individual contributor, as a small fry, you are kind of like, so what you are telling me is the CEO needs to come in and mandate it to happen.

Kind of like — and again, I haven’t really gone to DevOps, but in agile world it’s sort of like, if you are telling me I need to change the culture that means, A, I am going to do a lot of work that I am not going to be compensated for. B, I am going to be doing a lot of work. C, I don’t have the power to change culture, like I am basically going to have to do something.


And so that’s kind of where my reaction with DevOps come from, and the more nuanced thing is, if culture means, now we are going to spend a lot of time talking about processes and procedures and best practices and how you effect changes and/or we are also going to say, yeah, like the whole organization has to kind of like change the way it thinks about running the business, just like any big organizational change, and that’s really difficult. Then that’s acceptable, definitely.

But I feel like if you just kind of say culture kind of offhandedly, that’s kind of a way of saying, screw you, it’s culture.

John Willis: Yeah.

Michael Coté: But like you were just walking through a bunch of things, that we are like the iceberg under the water.

John Willis: Yeah. So in other words, you are saying what I am saying is a bunch of shit, right?

Michael Coté: No, no, no. And to put it another way, whenever someone says that some sort of technology driven thing requires a cultural change, it just begs the question, okay, what is the way that we change the culture? I always try to avoid begging the question sort of statements.

John Willis: A couple of points I will make. One is, I think one of the things that I learned early on from Lloyd Taylor, who is a DevOp guy. He used to work with Google, he like ran Op for Google. Had some really interesting positions in the Valley. He says, and I firmly believe this, you don’t change culture, you change behavior.

Michael Coté: Right.

John Willis: So if you look at it from like how do you change behavior and what are the things that can force behavior change? One technique I think is this idea of sense of urgency, like leadership trying to drive a sense of urgency. Hey, this project is urgent. By the way, wake up everybody, we are getting our ass kicked for the first time ever in this division.

I think the do now, ask forgiveness later behavior change works. We have seen — in the cloud we have seen classic examples of that. The Best Buy Blue Shirt Nation is a great do now, ask forgiveness later. Those guys went and changed — and they changed the culture. I mean, they put in a Drupal website that made it available to all the guys.

Michael Coté: And this is, correct me if I am wrong, but basically there was a bunch of employee or some employees at Best Buy and they have essentially set up their own community site.

John Willis: Yeah, they got tired of all the storage doing the —

Michael Coté: To sort of swap best practices and —

John Willis: Yeah. And what turned out a year-and-a-half after, the board started looking — going to like, before we put this out, let’s check with the Blue Shirt Nation, see if they —

Michael Coté: Right, and nowadays this would be considered like, oh, this is Enterprise 2.0. I mean, this is like social business essentially.

John Willis: I talked to those guys. I saw an interview of them, I get confused. And they were asked like, what if you had to ask for corporate approval? They were like, never would have happened. The famous ‘New York Times’ story in the cloud where they did like four terabytes and whatnot, whatever it was, like that’s another do now, ask forgiveness later, right?

Michael Coté: Yeah.

John Willis: So there are two ways to approach it. They had phenomenal success, now the culture has changed in ‘New York Times’ that cloud is okay.

We have talked about stories where Tivoli self-service story I have told many times, where these guys just changed the cultural behavior of how monitoring and self-service went.

So I think that’s another approach is, take a project that’s so broken or everybody is so mad at, change it, and then if it works, they will come. The behavior — other groups would be like, wow, how comes those guys are doing so cheap so well?

Michael Coté: Getting into that ask for forgiveness later mentality, I mean, it gets back to what you were saying earlier. You first have to establish a sense of purpose, because you sort of have to know what you are going to do and then ask for forgiveness. Like you can’t sort of — if you are kind of a passive IT person who is just kind of like responding to service requests and responding to tickets, you are kind of — there is not really that opportunity to take a risk and do something. Like it’s almost as if you have to think like, oh, here is something that the business isn’t asking me for.

John Willis: Right.

Michael Coté: Or maybe the business could be asking you for something, and here is a nontraditional way that we could satisfy that request, but just like with you, I mean, the Blue Shirt thing is a good example. It’s almost as if like you have to find someone to come up with a project out of the external channels, just because you are not going to be given the opportunity to do that.

It’s almost as — I mean, this — and this gets to a bigger philosophic question of like, what the hell is the point of a big company if everyone is always telling you to subvert the big company, to do stuff in big company?

But I mean, it is as if — I mean, that’s another cultural thing is, you almost want somehow the big company to kind of have that mentality in its employees is, we are going to establish stable business processes that we know are reliable and make money and we don’t screw with. But there is also some percentage of whatever that we want you to experiment and try new things out, because that’s going to be where we are going to come up with new stuff.


John Willis: Well, I think that’s part of the culture too. I mean, there are a couple of other patterns that I was going to mention is, a pattern I see a lot, which seems to work is take an Ops guy and put him in the Dev team. So that’s another one where kind of forcing —

Michael Coté: What do you think of that Bandolera thing, just a little break? So this lady just walked by, and they sell these in Austin, and I think they make them here. It’s basically like a fanny pack that you wear like a Bandolera and it seems — here is the whole issue, we will get back to practices and everything, but this is the South by Southwest special edition.

The whole issue is, you are familiar with the merse, the man purse? I mean, that’s a great idea, like I don’t want to carry around a big gigantic — like I want a purse, purse is wonderful, but you can’t do that. I don’t know. I am not secure enough in my identity that I can carry around a merse.

John Willis: I have got a cure for that.

Michael Coté: What’s that?

John Willis: Flat 50 in short.

Michael Coté: So this Bandolera thing is kind of like, it’s another way to kind of go for that, and I was thinking about it. It seems — and the alternative is the journalist’s vest, the fishingman’s vest, but that’s out too. Basically, the middle ground I have found is a sports jacket. One of the main reasons I wear a sports jacket, other than it kind of like makes people like think that you are better, is it has a lot of pockets in it, you can put stuff in there. Anyways —

John Willis: I am like — my next life, definitely sports jackets, all the way, from 20 on, next life, about 20 I am going to have like a killer set of like those plaid ones and really —

Michael Coté: Oh yeah.

John Willis: Not fancy color ones, but the browner or dark black.

Michael Coté: I think some plaid, I think you would do good in a sports jacket. So I was interrupting about introducing practices and patterns.

John Willis: Well, it’s kind of crossing functions, like in other words, put an Ops guy in Dev and have him like on a dotted line, to like the VP of Engineering, and it causes some interesting changes. I have seen it.

So I think there are ways to kind of — innovative ways that force behavior changes and when it works, then it becomes a groundswell, right? So I think that’s —

Michael Coté: And similarly, in the agile world, I think at about the point where people were running around talking about culture, there was this deluge of books of patterns, like people love doing agile patterns and, again, they were bittersweet — there should be a word that’s bittersweet, but it’s like frustrating sweet, it was like frustrating sweet to read because you would read this and you were like, that is totally the way people need to operate and run.

But it’s also like la la fantasy, like how am I going to — it’s like, how do you go in and get people to care about their dev?

And then the other dead end answer, and this is like a retrospective, sort of like I bet this stuff will come up again, and it even came up during the Etsy thing, the other dead end reply after culture, and then the counter reply to the counter reply as well, you have got bigger problems. That’s the other one, that once I hear that I like go off the wall, because it’s like, you don’t get it, it’s not that I have bigger problems, it’s that those are the problems I have.

John Willis: Here’s the reality, and this is the theme that I have had since the beginning of this, which is, the technology and the opportunities for companies are that, the more you ignore this crap, the more likely you are going to go out of business. And I don’t care who you are, we talk about this all time, there’s a boat that’s going this way and a boat that’s going this way, and back in the Internet boom, it was just the kind of retailers that got sacked, but the guys who did insurance and banking and all those guys, they were like, oh, that could never happen to us, Barnes & Noble versus Amazon, no problem. But today it’s anybody.

I mean, any business is —

Michael Coté: So let me interject some ideas here. So one of the — we are going to record another episode with an IBMer who is coming along here, or at least I am, I don’t know if you are going to stick around, and we are going to talk about IBM Pulse, so we won’t get into that here. But one of the things that I was delighted to hear one of the Pulse guys talking about is they use this term technical debt, which is sort of, 29:40 that’s one of his big things. He has done very well capitalizing on it, if you will.

Anyway, I think you are brushing up against one of your favorite things that I don’t think really has a phrase, but it’s kind of — it’s almost the opposite of technical debt and it’s like —

John Willis: It’s a business debt.


Michael Coté: It’s stuff you are not taking advantage of that can make your business better. Whereas technical debt is like technical decisions you have made that are going to hold you back in the future, because you didn’t do the right thing.

If you can measure the second thing, I think technical debt plus the second thing would be a tremendous forcing effect.

John Willis: There is a debt there. Yeah, partly technical debt, partly business debt, whatever it is. So there are companies now that make so much money and the inefficiencies are like, well, yeah, we can change it, but the story was somebody tried to go into one of the large credit card companies and they just hold case for like cloud and agile operations and infrastructure, but it was very compelling, a lot of money being saved, like billions being saved.

They are like, yeah, that’s great, but I can get one of these genius kids here to shave something — do some algorithms on credit card interest rates and make hundred contacts in a day.

And I would say, well, yeah, I mean, that’s the problem that you run up against in these large companies is that in some case —

Michael Coté: IT is kind of insignificant.

John Willis: Right. Or the significance of changing it compared to like other things they can do to make money.

Michael Coté: As they would say, changes in IT don’t move the needle enough to motivate changing a company.

John Willis: Right. But I think that is a form of debt, and it goes back to that Barnes & Noble and Amazon, they woke up like day one, but everybody else was like, well, we don’t have to be efficient, because — you don’t say that, that way, but you take, for example, I haven’t really followed them too much, but Simple Bank. It’s a startup bank. You know what I mean? And sooner or later those inefficiencies, that debt will pile up, and assuming that it will always be this way, based on what we have seen over the last three to five years of technology changes, it’s a dangerous way to play this game.

Michael Coté: I mean, essentially, it’s that you are not fortifying yourself against disruption. You are not protecting yourself against disruption. Like someone is going to come along and disrupt your business model.

John Willis: And by then it’s too late. Then it’s like, how do you react to — then it might be too late to change the game.

Michael Coté: You don’t have to worry about making noise. This is our guest for the next episode. Yeah, and I guess that’s —

John Willis: I am glad you brought that up, because that’s something I have been — there was a discussion about this last week about the whole — the inefficiencies of the enterprise, but does it really matter, but it’s debt that’s piling up.

Michael Coté: Because I feel like that’s the point things come to when you start talking about it’s culture and whatever is — it’s essentially like, tell me the three graphs I can show the management team that tells them to give me control to make these changes.

One of them is like sort of, whatever, technical. The other one is maybe like this idea of whatever it is, business debt or disruption problems. And I don’t know what the third is. But you need sort of something that’s — if it’s going to be cultural change it’s kind of like, oh yeah, the way we are doing this is crap, even though we don’t really care and we need to do something that gives us the main stuff. And then after that it’s the details of —

John Willis: I think the only thing I see, obviously, how to solve the enterprise cruft, if you will, or whatever, I don’t know. It’s a hard problem. Even Ernest Mueller, who is going to be doing a lot of stuff with us this week, at National Instruments, it’s a hard problem. I mean, you have got years of silos and groups, but I think the only pattern that I see working reasonably well is this idea that the kind of siloed new group, you have got a new project that’s on the horizon, treat it the way Amazon and Google, you build a team.

I heard a phrase at one of the DevOps meeting where they said, products not projects. In the enterprise we tend to think of projects. We put a couple of guys on it. Then when it’s done they go off and do something else, and the project is — whereas, no, no, this is a product, and it lives or dies based on its success. You might spend the rest of your life on that product, or you may transfer to another product, but the group is — these silos and then drive those as purpose driven and —

Michael Coté: Yeah. To wrap up this short episode so we can jump to the next one.

John Willis: I have got to get my Bloody Mary too.

Michael Coté: That’s right. I think that’s a good metaphoric way of thinking about it is, maybe you should transition from service management to product management, and like if you can understand that shift, then that kind of encapsulates a lot of what all the cloud and DevOps people are talking about as far as owning the IT that you are delivering and working with Dev people.


And basically because the whole point is, the whole thing that you are always trying to get Ops people to do is have a stake in the business. Like not a stake —

John Willis: A purpose.

Michael Coté: Have a stake beyond getting bonuses. Yeah, a purpose, but be involved in it and own it. And if you are just managing a bunch of services, you don’t really have a stake in it if you are managing a product in the same way.

John Willis: Well, the thing I think is — another promising thing I have heard of is this idea like, a lot of what — if you listen to Etsy and then the Wealthfront guys, they attribute a lot of this like what they are doing to that kind of Lean Startup movement, which actually originally comes from Steve Blank’s ‘Four Steps to Epiphany’ and then Eric Ries, who is kind of a disciple of that.

Michael Coté: Yeah, they had a whole track about that yesterday, over at the AT&T Executive Center.

John Willis: Was it a track or was this —

Michael Coté: The hotel that dare not call itself a hotel.

John Willis: I thought that was just — kind of looked like more like a VC hoopla.

Michael Coté: Yeah, yeah, yeah.

John Willis: But anyway, so they all kind of — you can trace them back to that kind of Lean Startup. There really Eric Ries has driven pretty hard, but I think what I heard recently was this idea of maybe the Enterprise Lean Startup. Again, that same silo approach.

The next time the enterprise wants to try out a new product, if you will, start it as kind of a — follow all the traits of these really successful lean startup business and do that internally, funded, given like $2 million and cherry-pick a team, build a silo team. And I think make me king for a day, and you are a large enterprise and you ask me how do I even start to kick the ball for DevOps, I would say, that’s what I would do.

Michael Coté: It’s almost as if you want to go back to like 1983 and reread ‘In Search of Excellence’, basically the whole pitch of that book. And there is a great follow on to that book called ‘In Search of Stupidity’, that was written in the 90s. And the whole pitch of ‘In Search of Excellence’ was, I forget the dude’s name, but now he is one of these famous management consulting types, and he profiled a bunch of companies, and I think one of his major conclusions was that at these big companies, they would run, they were entrepreneurial. Like they would allow people to run little products and projects with DevOps. It made sense.

John Willis: You can’t — like if you look at — again, I am a fan boy of these Wealthfront guys and if you look at what they have done in probably less than two years, and probably less than a couple of million dollars, they are a mini E*TRADE.

Here is the thing, so we talk about how does the large companies change, Damon Edwards says this, and I love this, that E*TRADE woke up one day and when they found out that the crust that Ameritrade was charging their customers was less than their internal cost to process trade. Like that’s when you have to wake up and say, holy shit!

Michael Coté: That’s your business debt or whatever your concept is.

John Willis: Right. It’s like, that’s the day, like holy shit, what are we going to do?

Michael Coté: The disruption onslaught.

John Willis: You would rather be in a position where you are at least like we are running some projects and prototypes that — anyway, well, it’s —

Michael Coté: Yeah, I think we have actually come up with some interesting ideas instead of just talking a bunch of crap.

So we are just going to take a break so you can get your Bloody Mary and then we will see everyone in the next episode, which for us would just be a few minutes.

John Willis: Right.

Disclosure: see the RedMonk client list for clients mentioned.

Categories: Cloud, IT Management Podcast, Systems Management.

Tags: , , ,