Skip to content

Pouring new wine on old problems – IT Management & Cloud Podcast #75

Overcast in the backyard

We’re back to news and current topics this week: Tivoli buying BigFix, CMDB’s, dev/ops, and more.

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:

Show Notes

In this episode, John and I discuss:


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 the 1st of July, 2010, and this is the IT Management Cloud Podcast, Episode #75, and as always, this is one of your co-host, Michael Coté, available at HYPERLINK “”, and I am joined by the other co-host.

John M. Willis: This is John M. Willis at People Under Process.

Michael Coté: Yeah.

John M. Willis: No. Botchagalupe, HYPERLINK “mailto:[email protected][email protected].

Michael Coté: And you are also — you are broadcasting through a Snowball today, right?

John M. Willis: Yeah, I got a new — it’s called the Blue, I guess, right?

Michael Coté: The Blue Snowball.

John M. Willis: The Blue Snowball, but —

Michael Coté: Maybe I am making up this Snowball thing. I need to look up if I am actually saying the right thing. But it’s the microphone you are using.

John M. Willis: It shows up as Snowball as your device is, so you must be right.

Michael Coté: Would you call it a good microphone?

John M. Willis: Well, Michael, I think my voice projects very well with this new microphone. What do you think?

Michael Coté: And it’s like — it felt like an enclosed device, right? Like I remember looking at — I was actually looking at this yesterday, because my iPhone just up and died on me. It’s time for the iPhone segment of the show. My iPhone up and died on me. So I had to go to the — I almost said iStore John.

John M. Willis: The iStore.

Michael Coté: So I had to go to the iStore, as I was saying, and of course like — since this is the iPhone segment, the Apple segment of the podcast, can I just say, what is up with Apple, their retail stores? I was there Tuesday at 4 p.m. — no, Wednesday, it was Wednesday at 4 p.m. and the thing was packed, they were running like 10 minutes late.

John M. Willis: It’s crazy. It’s insane. You have to — it’s like getting kidding — I don’t know, it’s like on Saturday morning having to get cold meat or something, that you have to take a number and — you can’t just go ask somebody, somebody has got to come up with a clipboard, and oh, how can —

Michael Coté: Yeah. And the thing is like, it’s packed with Apple employees as well, right? They are just like — people love that place, man.

John M. Willis: It’s depressing, because you pretty much know you are going to, A, get treated like shit and you are going to love it, right? So it’s like one of these things where they got you. I mean, there is a bunch of Apple haters here just calling us nut, and saying, well, just buy a damn Google phone. But did you get iPhone 4?

Michael Coté: No, no, no, it’s just a 3GS. I mean — and I have to say, I always have lots of complaints about the Apple Store, like I find it really obnoxious that there is not just a checkout stand you can go to, because I am a tech person, I hate talking to people. And I don’t like hunting down somebody and be like, oh, hey, can I buy this? I would like to checkout, and then you have got to do that.

Then even when you go to check-in for your Genius Bar thing, there is not really like a sign that’s like, hey, moron, who pays a 20-50% premium for commodity hardware, here is where you are check-in. You like have to figure out that you go talk to someone to check-in.

Anyways, all that said, like when I did get paired with my Genius, she was a very nice lady and she was like, sure, it would be nice if this phone worked, wouldn’t it? We should fix that. And then she just goes back and gets a new phone and boom, like I was out of there, like really fast. She didn’t — she wasn’t like, did you drop this in the toilet, like they didn’t really check or anything like that, so it was great.

I mean, it’s that — they do have good, what would you call it, I mean, the overall customer service experience is brilliant, because it’s overwhelmed, right, but you get good results, I guess, is the point. It’s kind of like what you would expect from like buying a ThinkPad or dealing with like IBM system stuff or whatever, where you are going to pay a lot of money, but they are kind of — their whole thing is just to like make sure you have stuff that works, not really drill you on little things to save pennies here and pennies there, because you are not saving pennies here and pennies there.

John M. Willis: No, you are not saving anything. But I did find my killer, killer, killer app for the iPad.

Michael Coté: Oh, what is it?

John M. Willis: Well, it’s — I don’t have the iPad in front of me, so I am not going to name it. But what I want to do is I want to be able to give a presentation with my iPad. I told you this before, right? I want to be able to hold my iPad with a little — with my little Stylus and actually kind of do a presentation on kind of a whiteboard on the iPad and have it show up on the big screen.

Michael Coté: Oh yes.

John M. Willis: But do it kind of wirelessly. So there is actually — it was interesting, I couldn’t find anything, and just the other day I found one. It has got this little thing where if you are on the same network, you can actually set up a web page on like your laptop, which could be connected to a projector, and then you can stand near your iPad and kind of draw stick figure pictures and tell your story. So that’s cool.

Michael Coté: Hey, have you drawn a stick figure on a stick tricycle?


John M. Willis: No, I have not done that. In fact, I don’t even know how I would go about doing that.

Michael Coté: Well, what’s a good segue from the Apple segment John is, did you see that Tivoli bought BigFix today, or announced that they had bought BigFix?

John M. Willis: The BigFix —

Michael Coté: Because apparently — so BigFix is — and I am looking forward to John’s Tivoli history moments when we go over kind of Tivoli in this phase. But BigFix is basically — it’s like all sorts of successful product companies, they weasel out of like good categorization. But they are basically desktop management software that’s been applied to all sorts of other devices as well.

So you want to say which software you want to have on various desktops or devices or whatever, right? Your computer is hooked up to the network that you want to manage, the configuration on it, and then you can run compliance and security stuff and a little bit of asset management, all of that exciting stuff, which is interesting, because this is one of the — as we talked about a couple of episodes ago, this is one of those areas that I am kind of like, this is all like hand waving analyst theory stuff. But I am excited to see this category kind of go away in the sense, and take that money and use it on something else. That it would be great if people just did this kind of management on their own.

But in the meantime, it does seem like — from what — I don’t know BigFix extremely well, but from the little that I do know, it does seem like it kind of fills a pretty noticeable hole, if you look close enough at what Tivoli has, in the sense that they are not like — people don’t really associate sort of desktop management and endpoint management, as they would call it, with Tivoli too much.

John M. Willis: You know what BigFix does, is fills a sink hole.

Michael Coté: Hey yo!

John M. Willis: No, I mean, — it’s funny, BigFix has been kind of a pain in the butt for Tivoli configuration for years. They have lost a lot of business to BigFix over the years. In fact, I was surprised to see they listed they only had 700 customers. Actually I would have thought they had more than that, because I mean, I can name five or six companies, off the top of my head, that I know went from Tivoli configuration management completely over to BigFix.

Michael Coté: I think the other thing is, I don’t know this for sure, but I think they do some sort of like OEMing and also selling to MSPs and other people. So their customer stuff might be like, one of our customers is someone that goes and sells to like 200 other people or something like that.

Like this is another, to interrupt your John’s Tivoli history moment, if I remember that title correctly. But anyways, like I was looking through their press releases and they actually had some interesting cloud related deals, which — this is kind of like cloud retrofitting, but they had some deals with like BT and Mitsubishi and some other people, where BigFix was kind of selling its platform to some managed service providers, who were then using it as sort of like SaaS or kind of SaaS like ways of doing configuration and asset management, which is kind of interesting.

And actually, to go on, on this point, I was — nowadays, when IBM has analyst calls, which they do for pretty much all acquisition they do, I can’t think of when they haven’t done one, it’s kind of cool, they send out a Twitter hashtag to use, which is crazy. It’s like, these IBM guys are getting hipped to the social media and they kind of have for some time actually.

So I was just writing like, hey, I am listening to this call and I wonder if there is a SaaS angle? And the analyst relations lady was like, there is a question from Twitter, and I was like a little — I usually don’t ask any questions in these things, because the answers I get are usually not always extremely helpful.

But it was Al Zollar, the General Manager of Tivoli was saying that they didn’t have anything to say about a SaaS angle, but there was things they were working on in the future that — especially with managed service providers that they would use in a sort of SaaS like way he was seeming to imply.

John M. Willis: Well, here is the thing, right? I saw that in your blog article too. You wrote a good article on that whole BigFix analysis of the acquisition. So the reason I was calling it, they fixed a sink hole is that, this is an area that Tivoli has just not been able to nut over, in the configuration management space, is the patch management and the kind of Windows security compliance, like how to get that whole zero attack day crap, right?

So they have wasted a lot of money over the years trying to solve it, they don’t seem to solve it. They get it — kind of almost there, but it’s like — at one point they had like, I remember somebody trying to put together their patch management solution and it was like, you had to install like eight servers to get this patch management solution up; three WebSphere servers and a server for this, a server for that, just to get — something you get from Microsoft for like nothing, right? Or from SMS.


So BigFix was one of those companies that kind of nailed that piece, I think patch management, and partially the whole security compliance issue.

So yeah, IBM has just fumbled over the years on patch management. So I think as I was reading some of the article of yours, and a few others, that said, this is an additive to the Tivoli’s portfolio, I think it’s true, because it’s additive in a sense that Tivoli has just wasted a ridiculous amount of money trying to solve the patch management. Or going back and forth; at one moment they will say, we can’t do this, and then they punt and then they realize, they have to do it to maintain any persistence in that space. Then they try to do it, they screw it up, and so now they have actually kind of gone outside to solve this problem. So from that perspective, I think it works together.

But the problem they have is that — alright, so let me give you the quickie. You want to go through the history for sure?

Michael Coté: Yeah, yeah.

John M. Willis: So if you look at what Tivoli has done for configuration management, if you go all the way back to kind of the mid-90s, when they were still Tivoli, right, early 1990s, they were Tivoli out of Austin. They basically had about four core products; they had a monitoring solution, they had Tivoli Monitoring; they used to call it — DM was the name of the product, Distributed Monitoring, they had Tivoli Software Distribution, they had Tivoli User Management, and then they had the Tivoli Enterprise Console, and they had Tivoli Inventory, right? So which is really kind of coupled with the software distribution.

And the software distribution was really just very much the tool that they used to deploy the packaging of their products. So it was a very clever, way ahead of its time technology, but it was never really well understood. And the problem they had is, Tivoli, early days, really not until IBM acquired them, really didn’t get any real traction in the Windows world. Now, they had some things, but they were totally lacking in the Windows space. So when IBM bought them, they put a lot more money into dealing with Windows and —

Michael Coté: That’s kind of funny, isn’t it? Like even IT management startups nowadays, like Windows is kind of what you do after you get money. Like tooling up for Windows is kind of never like a first or even sort of second item on the list, if you will. It’s kind of like, well, now that we have done a few releases and been around a while and had some funding and whatever, we are adding in Windows stuff.

John M. Willis: Yeah. Well, and it was part of that kind of distribution, going from the glasshouse to distributed systems, right? It just started getting like all these Sun systems out there, and all these HP, and all these AIX systems just showing up all over the place for different applications, and there was no management tools at all.

I mean, you have got to give Tivoli credit, I mean, they actually — I mean, I don’t want to go on for hours, but what they did was revolutionary. They did two really fascinating things; one is they put a stake in the ground for a framework for systems management for distributed computing.

Now, they forgot about Windows, but it really wasn’t a problem at that time, the problem was distributed computing, with all these platforms and not enough expertise to manage hundreds, and in some cases, not quite thousands back then, but close to multi-hundreds of servers, where very few people had the expertise to run these things, right? And we talk about that problem now, it was worse then, right?

So they put a framework in place for managing and best practices of monitoring, and the thing was that — the other thing they did is, they really built a CORBA-based framework that was like phenomenally sophisticated. And this was pre-CORBA — it was like the initial Tivoli, the guys who were right in Tivoli were basically part of the CORBA standards and proposals, and so a lot of the Tivoli is like 0.9 CORBA-based version, right? Because they actually released Tivoli, their original framework, before the original standard for CORBA was released, or what they call CORBA 1.0. So they did those two great things.

But the thing was that software distribution product was like not ready for the, A, Windows, it wasn’t ready for the bandwidth requirements. Like when IBM bought Tivoli, you have got to remember, Tivoli was a $50 million software company with maybe 50 customers or something like that, right? And IBM buys them and within a year-and-a-half they go from $50 million to $1 billion in revenue, right?

This software was just not ready for the amount of demands that the world was putting on it. Companies like Chase Manhattan Bank were like, okay, we now want to do software distribution that’s pushed to 30,000 desktops.

Michael Coté: Right. And then it’s like days 123 of training. Have you heard about CORBA?


John M. Willis: Yeah, yeah, right. Not only that — so what they did is, they basically — part of the kind of money they had from IBM and all that, they basically rewrote and they called it Tivoli Configuration Manager and they punted the old original technology they had.

And at that time you had Marimba and you had Novadigm, which were really leading the pack to get ahead of the old Tivoli stuff. Like they kind of came out of like, yeah, we get it, we are going to focus on this only.

Marimba was actually ultimately sold to BMC and Novadigm was sold to HP, but they did Windows really well. They handled a lot of what were called bandwidth problems.

So Tivoli wrote another product to catch up with those guys and also — but here again, it was pretty early on stuff. I mean, so we are still talking 1999, 2000. And then he made a ton of money just selling the dog out of it. This is another mistake Tivoli has made over the years, is not reinvesting back in their products. They make a lot of money on the product and they don’t — they let it kind of sit for a while, and they did that with a couple of their products, and Tivoli Software Distribution was that.

But the thing is, almost all of their customers switched over to this Tivoli Configuration Manager; they had Package Management, they had some new things, and then they acquired ThinkDynamics in 2003, which was supposed to be the replacement for all this.

So ThinkDynamics came up with this Provisioning and Intelligent Orchestrator and all this stuff, and then they spent the next five years, and the thing that they didn’t get is — I think we have talked about this before, that Tivoli didn’t get what is that — they didn’t know whether they were desktop configuration management or a server configuration management or both.

So what then happened was, when they basically punted on the old configuration management, they took the ThinkDynamics technology and they basically ran with it five years, and literally, it took them five years to figure out that they couldn’t handle the scale of their customers for desktop management; that stuff worked great for server.

Michael Coté: Because you don’t really have to scale it essentially. I mean, there is less servers, whereas there is a whole lot of desktop versus servers, and there is more churning, you have to check them more often, and they are dirtier. It’s like dealing with a block full of trashcans, instead of a one dumpster to deal with.

John M. Willis: Yeah. I mean, there is — I mean, some organizations have a millions desktops. I mean, even a larger organization, some place like BOA probably has half a million of their desktops running around.

So when you start talking about managing all those across the board, I mean that — you have got to have a product that like really churns, and it seems like why this is –.

So then, it was funny, they think the ironic part of this is, about a year-and-a-half ago IBM stepped back and finally realized that, that TPM technology just couldn’t do the desktop requirements, worked fine for server, and they rewrote a new one, and they called it TEM, it’s Tivoli Endpoint Manager, which is the one you have been talking about a little bit over the last few year, and they announced it at Planet Tivoli, the Tivoli Conference. And a lot of that is a lot of investment, which is a lot of overlap with BigFix.

So in a lot of ways, I have been following the Tivoli Mailing List today, and there was like a lot of questions, and the experts, they are like, I don’t know, I don’t know what this means, because they have been like converting customers all over to this TEM over the last —

Michael Coté: Yeah. I need to go back and listen to the Q&A section of the analyst thing, because someone was asking Al Zollar about basically that question, like, so how does this fit in with what you had? And I was distracted when he was answering. I don’t know what else was happening. But then I remember tuning in when he was like, well, we have done some stuff that to be frank hasn’t really panned out.

I mean, there is no like, to the point of you being able to rattle off like the three or four things, it’s no big secret or anything, but it is like — I mean, I think this is — recently I have seen IBM do a few acquisitions here and there that’s basically like the, to use your word, this thing called filing, where they are kind of like, screw it, we are going to buy our way out of this problem.

They tend to buy companies that you can tell are very like focused and they do one narrow thing and they have got — I mean, they are successful in their narrowness of focus and they are not solution-oriented, as it were, they are just some technological thing. Like in the WebSphere brand, I don’t know for sure, but it seems like the CoreMatrix people they bought are kind of like this, like they just do web analytics, which is a big category in itself, but for IBM it’s pretty minor, and rather than doing that on their own and so forth and so on. So IBM has been doing some good looking technological buys recently.


The troubling thing which — it’s sort of like the armchair thing that always existed, which is like, why can’t you invent this on your own? Which is like — it’s kind of like frustrating, but as your tone is kind of alluding to it, it is kind of like, well, that just doesn’t really work out that way.

John M. Willis: But here is a thing. So here is — I will put my future hat on and my experience with Tivoli. So I mean, if the rumor is right — like one of the article said they paid $400 million. $400 million is like freaking pennies. It’s the change that I let the kids pick up around the house and say they can keep it if they find it.

Michael Coté: Is that what you guys bat around when you are getting coffee there at Opscode, you are like, $400 million is nothing baby, we are going big?

John M. Willis: Yeah, that’s my budget. No. As I fly coach next week to somewhere, an unforeseen territory.

Michael Coté: Rough life!

John M. Willis: No, I got Platinum already.

Michael Coté: Yeah, definitely.

John M. Willis: But here is the thing. So like $400 million is nothing. It’s an area that they have like really failed out miserably. To put it the strong way that I wouldn’t say it is, they just failed miserably with the patch management.

So this is a technology that has gotten it right over the years in this space, and for $400 million, the question now is — the problem that you have with these technologies is people are looking for common ways to do things. We are sick and tired of having 15 ways to solve the same problem.

I mean, this is why Puppet, Chef, and these — we have such popularity in some of these Open Source tools is because, we are building abstractions around, okay, we are going to limit the ways you can do a lot of different things.

So Tivoli, unfortunately, is still following this kind of paradigm of, if you look at configuration management in the Tivoli space, for server management, you are going to have two or three different ways to do it. In desktop configuration, you are going to have the BigFix, some of the legacy, and then quite frankly, I don’t know that I would be buying ten year old technology right now. I think there is just too much exciting technology out there.

In other words, if this is either a throwaway and it’s a finger-in-the-dyke fix, then that’s great and $400 million doesn’t matter, and they can wait — I saw that, what’s his face, Sam Palmisano was quoted, he is going to spend $20 billion between now and 2015 in software acquisitions. So they are going to buy —

Michael Coté: Yeah. I mean, there was a — I will have to put a link to this, but there was a great article at the end of last year that was like a survey of cash on hand of big tech companies. It was kind of like — it was like shocking in that way, where it’s like, holy crap, people like Cisco and Microsoft and IBM and Oracle and all these other people, they just have like pile of cash sitting around. And I mean, obviously it’s nice to have piles of cash, but it also points to, there is plenty of cash for acquisition, I mean they don’t even have to get financing if they want to.

John M. Willis: So on paper, $400 million, I am sure this company will bring in IBM over the next two or three years, probably five years, way more than that, just from their customer base. They will get value out of it. It’s a great acquisition. But it doesn’t make their customer’s life in the long-term better. I think the answer is no. You know what I mean?

Michael Coté: I mean, along those — as everyone knows, I am not like a numbers sort of analyst or whatever, but I remember someone telling me that the Microsoft Configuration Manager or SMS as it used to be called, like the Microsoft thing that does all this, that it was basically like — and I might be remembering this wrong, so this could be totally wrong, but they were saying it was basically about a billion dollar business or something like that. There was a lot of money in it. And even if it wasn’t a billion dollars, it was a ton of money. So from that perspective, if they can eat away at that and other area, then it will at least pay off the investment.

But I think you are right in the sense that, of course you are in way biased of this view nowadays, but to be fair, we have talked about this stuff for a long time. So anyway, I am just making your chain, but there is this new — it’s almost like there is an Open Source approach to configuration management, which is, things like — like you guys in Puppet do, which is getting to that exact point of like, why are we sort of duplicating all this effort configuring thing — like coming up with the way we configure and manage things? I mean, that’s a very high level way of putting it.


John M. Willis: And the thing is, I think what we are learning is, is there is a collaborative approach. And what does collaboration mean? That we have this ability to kind of zero in on lots of people doing it kind of in a similar way. And again, I think this is an area where IBM — again, on paper this is a great deal. They are going to make money. Yes, I do have a dog in this race, so I am biased. But at the end of the day, I will still pose the question, whether it’s Chef, Puppet, versus anything they have, does this acquisition make their customers’ lives better?

Well, maybe it’s better, maybe it does. Maybe solving the patch management in some of their larger companies today is a solution, but does it make their job easier? And the answer is, I can tell you emphatically, no. In fact, it will make their life more complicated, because now they have got to decide, okay, do I stay with the original TCM? I have been halfway migrated to this TEM. Now I have got this BigFix. IBM is not really going to tell me for sure what I should do, and this is not anything bad on them, every vendor will do this. They will spend the next six months to a year figuring out themselves which is the technology that will win, which parts will win.

So they are not going to say anything about, oh, here is what you need to do today, you need to drop all that and go ahead on this. But in six months from now they will come out with their releases and say, hey, you have got about six months to do that. You know what I mean? So there will be a lot of confusion in their customer base.

Again, they are throwing ten year old, not new world technologies at their customers, which is not what they should be doing, they should be —

Michael Coté: Yeah. That’s got to be wrong, that, that Configuration Manager is like a billion dollar thing. I need to go look up where I got that from, because that sounds ludicrous now that I think about it.

John M. Willis: Well, I can easily see that.

Michael Coté: I mean, that’s just a lot of money.

John M. Willis: Yeah, but there is a lot of Windows — I mean, there is a lot of Windows production servers and once you get those things patched within a really short window, that zero, I mean, there are companies –.

We were actually approached, and I don’t remember the company, there was a guy, who actually was one of the original Microsoft developers of the patch management solutions that Microsoft use, he went out and created this spinoff company. They approached my old consulting company, if we would co-develop Tivoli Patch Manager based on their solution. God, I forget the name, it was a really well-known solution for patch management, but they had no framework, it was just standalone agent without any connectivity to the server infrastructure, other than the patch management going on. And they wanted us to build it into the Tivoli framework and allow it to just become pervasive as a patch management solution.

We spent about six months actually developing this solution. But the problem we ran into, the IBM, they are like, we are going to have one in the next release and it would be free, instead of paying these guys.

But yeah, as we went through that, as we talked to customers, the problem was real. The frustration was, we would talk to customers and they would say, oh, John, you can’t imagine the hell we go through on a zero attack. We have got to get 4,000 servers patched in like less than 12 hours. Everybody — it’s all hands. I mean, our software that we use, Tivoli Configuration Management, but it doesn’t quite work, so we have got special packages and patches and get them built. God! If you could have an automatic system that can really just pull it down and let it rip across our enterprise, and we felt comfortable in the policy and the security and the deployment decisions, we would do that in a minute.

Michael Coté: See, that’s the thing, it’s like, so annoying that companies have to deal with that. You should just like not even have to worry about that crap.

John M. Willis: Yeah, but here is the thing. So we were actually proposing a solution. It was kind of interesting, there were a couple of times in my old company where we actually tried to build a software product on top of Tivoli to standardize the solution. We would go all the way up to chain, everybody would love it, it was just great, and the management would say, well, why don’t we get that from Tivoli, why do we pay $10 million for their software and they don’t do that? We are like, they don’t. They are like, well, why can’t we make them do that? We are like, yeah, you can make them do it, but it will still take another year, and even then you probably won’t get it.

But we would lose deals. It would be like, you would be burning in hell, it would be horrible, but this idea of like, I already paid $10 million to Tivoli, I am not paying another $100,000 for anything. Well, but you will fix it. I don’t care. I don’t care that it’s broke. I paid $10 million, I want them to fix it. Well, they are not going to.

The discussions you would have with these guys would be just mind-boggling, like, well, it’s broke, you are burning, the world is dying. You pay $10 million and for another $100,000 you are not willing to fix it. Why? Because they should do it, not you. Alright.


Michael Coté: Well, speaking — while we are on the topic of good old fashioned IT management, Big 4 stuff, I have to admit, I haven’t like looked into this too much to really do more than just kind of read what’s on the tin, but I saw that HP released a new version of its sort of suite of Management Software, which I think goes under the monicker of Business Service Management, somewhat quaintly.

So they have like their HP BSM 9 out nowadays. And it has a little bit of — supposedly it has some cloud stuff sown into it, and all the usual suite of stuff. But like I said, I have to admit, I haven’t really been able to get my brain around it too much.

I went to their web page to kind of look it up to kind of just basically get the rundown. And I had a tweet, where I was like, oh, I hate web pages like this, like you go to the web page and — most everyone who is listening has probably had this experience many times. When you go to your web page and there is like 30 things, 30 pieces of information, but nothing that really is sort of just like, here is all the features that it does.

And to be fair, there was a PDF here, there, that had like a little layered cake or a burger of everything, but it’s just — man, it starts to get annoying when systems like this, that are pretty technological, are so dense to just figure out what the hell is actually going on.

John M. Willis: You love those burger diagrams, don’t you?

Michael Coté: Yeah. When they are good, they are good, but they can be kind of dry and not very good otherwise, so it’s tough, it’s a tough thing to figure out.

John M. Willis: Yeah, I didn’t follow any of that. They are all doing it wrong, Michael.

Michael Coté: It’s like a canned sound bite you are going to have, press a button.

John M. Willis: Every time you mention, I am like, yes, they are just doing it wrong.

Michael Coté: But that gets to a broad topic. I haven’t finished reading the kind of roundup, because I stopped reading to go get a cup of coffee before the podcast. But I have been noticing a little bit of discussion about what the hell is up with CMDBs nowadays, and not in the sense of like, what is up with them in their self, but like how do they fit into the broad world of — the whacky crazy world of cloud stuff and virtualization and dynamic things?

And our friend William over at Oracle had a — he had a lengthy like response to someone who had written, basically it was like CMDBs don’t work for the cloud and things like that, and he was addressing some of the points, that some were spot on William was saying and others weren’t and so forth. But it is — it was just one interesting point, because I have actually had this discussion with several people recently, who are trying to figure out — they are trying to figure out like what to do about a CMDB or an asset database or all this stuff of just keeping track of things in your IT and the new ways of doing things.

John M. Willis: I didn’t get a chance to read it, but I mean, is his point that a CMDB doesn’t work because all the resources can be ephemeral?

Michael Coté: So William is actually pointing to this other guy, who is Bernd Harzog, and I think it is Bernd, because it’s spelled B-E-R-N-D —

John M. Willis: I went to high school with that guy.

Michael Coté: Being a U.S. English speaker, when there is three consonants, three or more consonants together, I don’t know how to pronounce it, I need a vowel somewhere. So I am just going to say Bernd.

Anyways, so here is his six points, I will summarize them for you. And this is a summarization of William’s summarization. So let’s see. So you have got a whole bunch of virtualized stuff, and let’s see, and then the virtualized thing — so you have got this whole new way of — you have got a new topology of virtualized things, and then the guest on the virtualized stuff. So that’s the first challenge, right?

So you have got another layer of complexity and so instead of just being physical and application, you have got that, so that’s an issue.

And then there is this whole new set of relationships between those things and then also hypervisors and virtual networks and virtual storage and everything. Seem to be he is not able to — weren’t designed to handle that.

And then you have got stuff just being created at a rapid rate that’s overwhelming to CMDB, this guy is saying.

And then also, the environment can change, so you have got VMotion — this is the VMotion dynamic thing. So things are moving around and then you have got, let’s see — and then you are also potentially splitting your assets between public clouds and private clouds and fat-free clouds and all sorts of weird clouds with adjectives in front of them. So it’s not just on your own network.

And then this is more of just a complaint about CMDBs in general it seems. But for a CMDB to be useful it needs to track all sorts of other stuff, like performance and service, SLAs and all sorts of things like that.


And what I like about this is, this is a good summing up of the — like I was saying, the discussions I have been having with other people who are either working on this or they are using systems and they are kind of like figuring out, I mean this is like — this is the process of, we reinvent the same shit every ten years. So this is a cynical way of looking at it. Essentially, we are going to reinvent the data store that you put all this information in. The old way doesn’t work, and the new way we want to do things, so we need to reinvent it.

But it is — you know, John, it’s never good to pass up on — you always want to take up an opportunity to beat the CMDB.

John M. Willis: No doubt. No doubt. I am going to defend the CMDB.

Michael Coté: What? Everyone is doing it right, is that going to be the sound bite?

John M. Willis: So I am the kind of guy like, when everybody is buying a Harley-Davidson, I think it’s time to buy a Kawasaki. But the thing is — so I was wondering — it’s good you summarized what they were saying about — what William — Bernd, my friend from high school’s overviews of why CMDB suck, right?

But here is the thing, I hate the like — each cloud is different, because that’s a bunch of poppycock, right? I mean, this is still data center 101, right?

I mean, if you tell me — and that’s part of those things you were just saying, like if you tell me that the rate of change is probably the classic way we structure the CMDB is probably not going to work well, I agree. If you tell me that — VMotion I think is poppycock — well, I shouldn’t say this, lot of people like it, I mean, I think VMotion is a silly idea in today’s world, right? If you need a new resource, you just spin it up. You know what I mean? You just spin up.

It’s like the idea of — I am getting off on a tangent here, but it’s the idea of the server being important, and so important that if something goes wrong with it, I have got to slowly migrate it over to somewhere else. But that’s not the way you should be designing architecture.

Michael Coté: Right. You don’t want like packrat mentality with infrastructure.

John M. Willis: Right. You have five of them and if one dies, you start up another one, and it’s all decoupled. You don’t have this kind of dependence or couple dependence on the server, and my God, if the server dies. I mean, there are a lot of legacy systems like that, and God bless the people who have to manage them, but the bottom line is this whole idea of VMotion and having to like keep the routine state moving.

I mean, again, if you look at modern architectures today in the cloud, it’s about decoupled and multiple servers that get spun up, one breaks, you add five. In fact, you listen to a lot of these kind of DevOps stories now, where people are actually — I mean, they will have a whole segment of their customer base that are not even seeing this new feature. So it’s even beyond like the server, it’s like — there is like switches and knobs throughout this mesh of application component tree, that turns on and off things depending on who you are as a customer. And that’s how people launch new features or turn off new features if don’t work.

Michael Coté: Right. I mean, I think to summarize it, you start treating a server as a component of an application and it’s a replaceable component essentially. So if something goes wrong with it, you don’t need to preserve the server, if the application you are having to support allows for this, which is where the whole legacy brownfield comes in.

But the better way of doing things is to have an application that supports treating a server as a component, such that if something goes wrong with the server or you get an HA situation, you just bring up a component somewhere else. And of course there are all sorts of like data management transactional stuff to worry about, but the point is that, that somehow the application should be taking care of that, not necessarily it be bound to a server.

John M. Willis: Yeah. I went off on a tangent there, but I mean the thing is — but if you are talking about the ephemeral nature of, things will start off and go away and the trend might start up, and before you even have time to capture that they existed, they might be gone. And in that case, yeah, it’s going to be hard to have a static CMDB representation of those resources. You know what I mean? And today, I think the CI structure of a CMDB is that, you have the assumption that you have the static architecture throughout, right?

Michael Coté: Right.

John M. Willis: So I think it has to be — but again, I hate this, like, well, the cloud makes a difference therefore, because there are a lot of things about the cloud and this data center 101. So that’s my defensive to CMDB.


But I think our good friend, Doug Mcllroy, I mean, if you remember 40:03, he talked about an ESM DB, he had actually done something like that back when he was at MindSpring. He was kind of explaining that you needed an extension to CMDB, because — for the management.

Michael Coté: I think another aspect is — I mean, there is, probably for many CMDBs, if not all of them, I am not intimately familiar enough with CMDBs to know if this is the case, but I would guess many of them just don’t have the technology, if you will, they don’t have the features to know how to do querying for discovery and finding out cloud-based stuff, right? I mean, that’s another thread of this conversation, and William hits up on this a little bit.

But like I have had this conversation with like — with a couple of people, like Zenoss came out with a new thing that — their 3.0 version is supposed to manage dynamic stuff better, and there is other people I have talked with. And it gets back to this thing that we mention every now and then, is that basically like, when you get into these cloud-based scenarios, there is all of these services you take for granted as existing in IT management that don’t really exist in a reliable way in the cloud world.

So you can’t like — you can’t do like SNMP scanning to discover a bunch of stuff necessarily. So that’s another part of the issue. It’s not so much like the CMDB — it’s the CMDB versions that are out there don’t support it, but the concept doesn’t necessarily preclude it.

John M. Willis: Yeah. And like if you look at like — so Adrian Cole works with me now, he worked with state clouds and you should see the hoops he has got. We all talk about, oh, standards for this. He goes to a lot of hoops to abstract just two primitives for clouds. I mean, JCloud in all its popularity, and there is something called FOG, and there is a couple of others out there.

In the case of JCloud, for all its popularity, it’s really just a two primitive; and you can shoot me and yell at me if I am wrong here, but it’s just like compute and blob, storage blob, you know what I mean, and it’s just a pain in the ass just to build that over about five or six different clouds. Just to do that alone has been a massive project with a lot of contributors.

So yeah, somebody having the push button way to interact with all the different clouds; private, PaaS, SaaS, for resource CMDBs, it’s going to be virtually impossible, but here, you are going to hear it here.

So here is my crystal ball, and again, coming from the dog in the race guy, so a horse in the race guy is dead, we are going to have to see is, and I truly believe in this, is the, the infrastructure is code model, so we have to start treating the infrastructure more like what we treat our code base.

In fact, there is a company, that’s one of our users, that — and I might have mentioned this, at the end of every sprint, they not only rebuild the code base, but they rebuild the infrastructure. So every two weeks at the end of their sprint, they actually start fresh with everything.

And the reason they can do that is they have their infrastructure as code in a source repository, in the same repository as their development, the application code. So then what you will have is you will have an operation repository, which includes all the definitions that build your infrastructure, and part of that will be like any application, will be dynamic things that they can grow and shrink.

So your code, your static code for an application, like the Java app, when you look at the code, you understand that there are things that once it gets deployed and it’s executing, it’s not going to look exactly how the code is, dynamic things happen.

Well, that’s the same thing as what — as people move, I think, and accept this, unless somebody comes out with a better idea, I think infrastructure is totally the right way to do this, that it will be very much. So you will know what your operational aspects; you will have an operation source repository to basically look at, again, things like Chef and Puppet to say, I pretty much know what my infrastructure should look like at any given time. And then what you are going to need is the ability to have a dynamic inventory of those resources.

So people are going to have to really rely more on the components, if you will, phoning home and saying, hey, this is what I look like. This box came up as a Ubuntu and here are the things that are on it. I have discovered this. Go into the inventory, and infrastructure of code told me I should actually extend onto that and become an Apache server with this extra plug-in and all this. And by the way, send that back to the dynamic repository, and if Autoscan detects that there are five more that should be built because of some scaling issue, then those will go to the repository and then if two of them go away, they will come out of their repository.


So I think that what we are going to see is a lot of thinking about those two; what it’s going to exactly look like, is ITIL going to be part of it? I think it really should. That’s probably an ITIL version. And hearing a million people — we don’t have a million listeners, but as many listeners as we have, we just heard a big —

Michael Coté: That’s right. As people like to joke, The Listener.

John M. Willis: Right. So yeah, I don’t know, I mean there is a lot out there.

Michael Coté: And so — like you keep using the word repository, I mean is that like — is that how people are thinking about in this space? I mean, the place where — the thing that keeps track of what the state of everything is?

I guess it’s not necessarily even — it’s not the state of everything in the sense of like processor load and storage, it’s not the monitoring, the performance matrix, it’s more just like —

John M. Willis: So it’s the abstraction of the definition of your infrastructure. It’s some type of representation. And if you hear anybody from kind of the DevOps movement or somebody who is using Cfengine, Puppet or Chef, they will tell you flat out that those definitions that describe the infrastructure should be in a source control repository; preferably the same one that your applications are in. You know what I mean?

Michael Coté: Yeah, yeah. I mean definitely, like there is the — I mean, if you have built everything up from a model-driven like automation or configuration management, then you obviously have those models of what your desired state of everything was, right?

And then to a large extent, you also have the relationships between things, because in theory you have built up your infrastructure in a way that its as a whole instead of a bunch of independent parts, so the relationships are part of that.

And then there is just the challenge of tracking drift and tracking things that change and like how things have evolved, so updating those models. I mean that becomes more difficult, and then having a nice interface on top of it.

I mean, I guess there is probably many people who would be happy to like checkout like their CMDB out of Subversion or Git, or whatever wacky thing they are using, but there is other — I think there are some UIs on top of it.

John M. Willis: Yeah. I think in a lot of ways the CMDB becomes kind of the idea of a static model that defines I think for the reasons — again, I defended parts of the reasons that I hate to hear that CMDB sucks, and that’s because, oh, it doesn’t work with the cloud.

I mean, it doesn’t work with the cloud, because, like we said, standard APIs with different cloud abstractions, it doesn’t work because cloud supports this ephemeral nature of components coming and going and existing for short intervals.

But I think in general they are right, I think the CMDB is something that needs to be completely torn down and rebuilt. I think part of it is probably — the problem is, is that you have this kind of, like I am sitting at Velocity and John Allspaw is giving a presentation, and he threatens to throwup a supplementary slide on ITIL and 300 people in the room go, ah, gank, and these are the WebOps like guys that are making it happen for the phenomenal web operations and new infrastructure, and they think that that the word ITIL is a dirty word, right? So I don’t know how you get over that hurdle too.

Michael Coté: Yeah, it’s like telling a cowboy to trade in his horse for a jeep, because it’s more efficient to get the product to market and then they need less cowboys.

John M. Willis: Yeah, and then they have got this like, my God, that’s — we are not going backwards here of course, no way. But the truth of the matter is, if you listen — I actually had a chance to talk to John Allspaw. He gave his presentation and I was listening to it, and he says — John Allspaw, actually, he has written some of the — he just actually was the author of this new WebOps. Have you got your — did O’Reilly’s send you a copy of the Web Operations book?

Michael Coté: No, no, I need to go get it.

John M. Willis: You need to get your copy. Can you get two? But it’s — yeah, I had a chance — but I had a chance to look at Jesse Robbins and John Allspaw, and just all the thought leaders that you would now call WebOps guys, or each to each. It’s kind of like Beautiful Code for WebOps, that kind of — where each guy wrote a chapter on different things.

Bu, John Allspaw, I talked to him, the thing he talks about, you have got to have change management, you have got to do incident management. And I am thinking like, somebody from BOA coming in for the middle of that session and going, these Yahoos, I heard how smart they are, and they don’t — they are just learning now that you need change management. We have been doing that for 20 years.


But at the same token, if John Allspaw walked into a Tivoli session and they saw our screen that was showing block and tackle file monitoring, disk space monitoring, and high CPU for 10 ticks in a row. John Allspaw is like, you are kidding me, take away my laptop. I basically on my laptop monitor all four walls 50:19 at any single time. You know what I mean? Everything mapped to the business of what they are doing. You know what I mean?

Michael Coté: Yeah, yeah, yeah. I think — definitely. And I think there is that — someone has got to have to be able to do — to bridge that ITIL — the process gap. Because there is definitely, as you were alluding to, you have got to have problem management and incident management, there is definitely a process that people do. And it’s either the case that no one wants to talk about it because it’s so obvious that it’s boring, right? Like you don’t talk about, part of my job is to get up in the morning and drive to work, I am going to spend three slides on that, right?

Like either it’s some trivial thing like that, or the community, as it is, unfairly not wanting to talk about it, right? I think somehow the Agile software development people figured out how to make people ludicrously excited about talking to me about process. So I am pretty sure it can be done in the operations area.

But I mean, getting back to the CMDB thing, I mean it sounds like what would be — essentially, it would be nice to have sort of — to mix together logs together in the way that like Splunk and other like crazy log people do, and use a bunch of the big data analytic stuff out there. We used to talk about magic cubes of analytics about a year or so ago, when that was like an exciting thing, and applying that to IT management.

But it seems like that magic voodoo could be applied really well of, taking these desired models, if you will, and doing like a snapshot at whatever interval. So you have an idea historically of what something looks like. And if you can kind of sort of like pivot, if you will, pivot around all that logged data, then it would seem to give you that historical CMDB data that’s really important for big companies to have for compliance and troubleshooting and all that kind of business.

I mean that type of activity, if you will, I don’t really see a lot of that going on in sort of cloud stuff, like root cause analysis and troubleshooting and things like that. I mean, I think that’s where there’s a lot of — there is a lot of things good old-fashioned IT management can offer in that area. And there might need to be — and there is no doubt a lot of retooling or adding of features that needs to happen, but that’s the kind of stuff that, to your point of saying earlier, like that really never goes away.

I mean, assuming that stuff is still going to break in computers, you are going to need to be able to do a bunch of troubleshooting and things like that, which I don’t know, maybe if you are using a Platform-as-a-Service, these issues don’t come up as much, but we will see.

John M. Willis: Yeah. You know what, I will tell you, I think that the two things that I have taken away from last week with Velocity and then the DevOpsDays that we had at LinkedIn, is that, there are two things, and I heard over and over from different people is that — And I think Adam and Chris covered this on the podcast last week that we had, is that, WebOps is becoming the dominant operation model. It shouldn’t be called WebOps, it’s Biz Ops, it’s basically where the rubber meets the road in terms of IT infrastructure, and everybody want to be a WebOps.

So if you are Bank of America, either you don’t know it or you are going to figure it out — you have got to figure out how to become — and again, I think WebOps the one versus Biz Ops.

So I think that I get to define — I am going to rename WebOps to Biz Ops and I will tell you that, what I define as Biz Ops is the dominant operating model moving forward. So that’s one thing I think that is clear; you either get that or you don’t.

And then behind that I think is, DevOps is a powerful movement about to emerge. So there are a lot of people that are sitting in the 54:07 of really influential people that manage really influential shops, that are coming to the table to say, let’s figure out how to do this right now.

In other words, as strong as the Agile movement was in development, we have got that kind of groundswell happening in operations right now, and good name or bad name, it’s called DevOps.

So the point is, I think we see, unless this thing just fizzles, there is this opportunity in the DevOps movement to get things like what it all should be. Should it be part of the conversation, shouldn’t it be? What is a CMDB in this new world? Should there be one? Should it be refactored? I think there is a phenomenal amount of opportunity for thought leaders to drive something that’s just — the operations guys, they are coming out of the woodwork, like, yeah, we had always thought we were important. I am surprised now everybody recognizes that. You know what I mean?


Michael Coté: That’s right. Well, the car mechanic is certainly important when you need to fix your car, otherwise — I mean, the whole point of a car is, you don’t have to have someone to manage it, which I think is —

John M. Willis: Unless everything becomes Indianapolis 500, then the car mechanics are the most important guys around.

Michael Coté: Well, now that you have brought up this spectra of DevOps, John, it’s a good chance for you to summarize how DevOpsDays was last week. Because the last time we recorded, it was a Thursday and DevOpsDays was on a Friday. So I know you and Damon talked about it on the DevOps podcast that you guys do, but give us a little summary over here, how was it?

John M. Willis: Yeah, it was really great. We had north of 250 people there. It was a free event. It was at LinkedIn. They provided a great facility. It was a big old room, theater style room.

Michael Coté: Hey, did LinkedIn go over like what their infrastructure is like? I ask because someone was talking about it with me earlier and there was something interesting about it, but I forget what was interesting.

John M. Willis: No, they didn’t. I think one of their guys was on the panel, but it was more — it was about —

Michael Coté: Like do they do stuff in Python or something? Maybe that’s what it was. I was talking with someone about like Python recently, and kind of saying, well, there is not a ton of Python people out there. If you are hiring Python people, it’s kind of best to hire a programmer and teach them how to do Python. If you are lucky enough to find someone who is like a Python programmer, that’s fantastic, but it’s not as easy as just like scooping up Java and PHP and .NET people. But who knows?

John M. Willis: Yeah. Well, that’s another whole discussion, but I think that kids come out of college that are actually winding up as sysadmins, so there is this kind of, Python might be the next Perl. You know what I mean?

Michael Coté: Oh! Exciting! But I interrupted you. So you were at LinkedIn there.

John M. Willis: So the whole day was fantastic, because we had these sessions on how to use Python, how to use Python in the clouds. There was just everything you wanted to know about Python. It was a phenomenal —

Michael Coté: That’s right.

John M. Willis: You like threw me off course, Michael. No, it was — actually, I have got to give Damon credit. It was a lot of discussion on how to do this open session. I was in a lot of the early planning things. I kind of jumped out and let Andrew and Patrick Debois, Andrew Shafer and Patrick Debois —

Michael Coté: Yeah, I think you and I are good at that, like weaseling out of actually doing anything when it comes to events. That’s a core competency that you and I share.

John M. Willis: I didn’t get much credit for it, but I was there. I got to sit at the control booth table, like nobody can push me away. So I think that was my little perk for at least getting the thing rolling.

So early on there was discussion of — so Damon felt really strongly that one of the problems with open sessions is that, it’s great, but you really want people to kind of prepare, because you lose some value. I mean, there is greatness about BarCamp and open session, where people just get up, but sometimes you have some really interesting people that aren’t like road warriors like us. You know what I mean?

Michael Coté: Yeah.

John M. Willis: That they might be there, but if you ask them to get up and give a presentation, they don’t have a stack of presentations on their laptop. They really need to sit down — like they work for a company like Ning, or they work for some big company and they are not really somebody who is presenting all the time. So he kind of drove that with a little bit of resistance of that, I really want a format that promotes open spaces, but also promotes this idea that — like one of his complaints about OpsCamp was, the only people that actually present are people like me and Anna and Luke and people that have the kind of handbag of all their presentations.

So if we really want to get the doers — I am not the doer, but the guys that are working in the trenches, we really need to give them some time to show up.

So the format was, these panels, it was all day panels with kind of a panel open session. So the assumption was that half the panel would be kind of moderated and the other half will be the audience kind of opening questions. And it really, really worked out well. I mean, a lot of the comments were that almost anybody in the audience could have been on any panel, and in a lot of cases there were lines of people by the microphone asking questions and starting discussions with people in the audience.


Michael Coté: Oh! That sounds fantastic!

John M. Willis: It really was. It was like, again, it was 250 people like extremely passionate about operational infrastructure or what we call DevOps. We had subjects on — you know a lot of the things came out, like a lot of discussions were about the culture, just a lot of discussions about culture in the data center, which a lot of people will talk about the DevOps being that, breaking down the wall between developers and operations.

So there was a lot of discussion about building culture and adapting culture. So whole sessions on — there was one guy, this guy was an investor and he was brilliant. He was like, one of the ways to solve — your buddy, Israel Gat, was there, he was brilliant too. So he was on a couple of panels. He was on a panel with this guy who used to work at Google and he said, you want to break down the wall, he says, just simple things. Put a picture of your family in your cubical. I mean, it just gets you one step personal to somebody. You know what I mean? Like oh, your kid plays soccer, wow.

And he is like, take out somebody you don’t get along with for lunch and talk about something completely — find something that you both are interested in and have nothing to do with work.

So there was a lot of discussion about this idea of the culture and how do you promote the culture and how do you get people to think that we are all on the same team. It’s not like, Jody Mulkey who is a CIO of Shopzilla said something like, we try to get our people to make the problem the enemy, not the people, right?

Michael Coté: Yeah. That’s a interesting parenthetical thing that I came across last night, is that there was actually — in one of the little IT news things I read, it was like the Wall — there was this — we have talked about a couple of things on this before, but it’s like the Wall Street like Technology Journal or something, which — there is usually a bunch of boring banker news in there, but every now and then they have kind of a fascinating enterprise, the IT thing. And they pointed to this PricewaterhouseCoopers, I think it was like — like all big consulting houses, they have like all sorts of quarterly PDF journal things of various amplitudes of ponderousness.

But there was — sort of like the opening editorial thing was this commentary on how, supposedly, they have been talking with CIOs, and these CIOs were kind of like, this whole like IT as a service thing is kind of putting us at a disadvantage in financial institutions, where they look at us as a utility. I mean, we have done all this work over the past ten years to turn ourselves into an utility basically, which means no one really collaborates with the utility to do work on core business things. Like to do things that are valuable to innovate.

And it was interesting — it was kind of — it was interesting in the aspect that you don’t really expect — I mean, that’s been the line for IT management forever, that you want to do IT as a service and service the business and all this. Whereas, they were kind of trying to break that down a little bit. To the point of what you are saying, it’s sort of like, you should be involved in the whole team effort, right? You shouldn’t be this weird separate bureau of shared services and things, because it sort of damages your standing in the company essentially.

John M. Willis: No, I agree. I think the beauty of it — so there was — I moderated a session on cloud and DevOps and there were sessions on monitoring. Our good buddy, Javier Soltero, we had some people drop out on the monitoring, and I knew he was in that area and so I invited him to be on the monitoring —

Michael Coté: Did he drive up in the cloud Carrera?

John M. Willis: Yeah. I didn’t happen to see his car, that’s a good question. I was with him all day long; me, right inside with this 1:03:48. I love that guy. But I was with him all day long about different things. But John, I needed you to help me. That ought to have been the ultimate one, oh, come on, please let me see your car, let me just take a spin at it. But he is just the guy now.

Michael Coté: Yeah, but of course.

John M. Willis: He was on the panel. He stole the show. You know what I mean? On DevOps, and he talked about the history of things, the Hyperic.

Michael Coté: Is he still using the phrase crapplication or something like that? I saw that crop up from a Gartner guy in some news report recently. When Javier was moderating a panel I was on at the Spring Conference last year, he was mentioning — he is like, there is all these little crapplications, like little crappy applications that you can run in the cloud and do stuff nowadays, and it was a nice way of putting it. There is like lots of little applications out there.

John M. Willis: Yeah. No, it was good though. It was good for him, like he was — even on the panel he was like, I am just really excited to see people — because he has kind of moved up into the big shop world now, and for him to kind of go back and just say, hey, this DevOps thing is crazy. It was a fascinating day.


Michael Coté: Yeah, it’s so weird when I talk with mainstream IT management people that they haven’t really heard of DevOps stuff. It’s shocking.

John M. Willis: I think it’s something that’s just going to start bulldozing. I mean, I have been calling this for — I mean, I was telling Israel Gat, my kind of first taste of this DevOps was your podcast with you and Andrew and Israel Gat where you were talking about Agile infrastructure. I was driving on the road, I pulled over, I was like, this is crazy, this is great! And I called Andrew, I am like, what is this, this is awesome, I love this idea.

And meanwhile, I have been listening to Adam Jacob and the Opscode guys talk about infrastructure as code, but this idea of Agile infrastructure, I was like — and he was like, well, there is a guy over in Europe called Patrick Debois, he is doing all this stuff over there.

And then I went to the DevOps conference out in Belgium, I am like, I have got to go here, and then we started a discussion about bringing it over here.

So I think the thing is, is all these 66:10 write is. One is, I want to write is, how can it be wrong when it feels so right? I mean, you don’t have to explain the concept of DevOps to anybody more than once. You know what I mean? You there?

Michael Coté: Was there any discussion of another DevOpsDays some time?

John M. Willis: Well, there is a whole bunch of regional ones kicking off. So there is a SoCal one in a few weeks, in Los Angeles, the London guys ran one last night; they are running like regular ones. I had like four companies out of Atlanta, I didn’t even know they were running in Atlanta, came up to me and said, hey John, can we get a DevOpsDay monthly meetup going in Atlanta? There’s guys already running a monthly one in Boston.

There is going to be a DevOps in Germany apparently at the end of the year. So I think we will probably piggyback. We will run a DevOps U.S. probably the week of Velocity, if that keeps working out, and then they will run a DevOps Europe, somewhere in like October, November time frame. And yeah, and I think it’s growing. They had a DevOps Conference in Australia down there. So we are all like trying to keep the thing connected too. So it’s — I mean, it’s a large group of people, but it’s a small group. So it has got — like they say, it has got legs.

Michael Coté: So is there like some website for all of that nowadays or like how is that being like managed?

John M. Willis: In fact, there actually was a discussion — there is a lot of people individually doing something, and they are like, are we ready for some kind of defining — I mean, it really is — I mean, I heard people talk about a DevOps manifesto and I kind of puked when I heard that the first time, and I still kind of puke, but I do think that there is — one of the things that somebody said the other day is that, we hate — it would really suck in a year for now we are still saying, DevOps, yahoo! DevOps, you know what I mean?

Michael Coté: Yeah.

John M. Willis: And it’s nothing different than we have right now, or we are still giving pitches, like what is DevOps?

Michael Coté: Yeah, yeah, yeah.

John M. Willis: So that has to happen. I mean, the DTO guys are on top of a lot of things, right? So something will come out of there. But I think there really needs to be some kind of — one of the things you, me and Damon were playing around with, we talked about this in our podcast, it’s like, kind of cloud has its SPI, and we were just kidding.

But there is three themes that come out of DevOps right now that I think are strong. One is culture, dealing with culture. What are the subcategories of what that means in a DevOps world? Automation clearly. I mean, we know that. Like the tool chain. Products like Chef, Zenoss, Puppet, all the different kind of — that you see in the ToolGen categories from DTO, something like that. And then Matrix, right?

So we were even kidding around like MAC, Matrix, Automation, and Culture. So that was kind of at least the first defining pillars.

Michael Coté: Yeah, define the category, that would be fun to do.

John M. Willis: Yeah, those are three — there is probably others, but those are three — if you go back and you start kind of listening to what everybody is saying, those are the three dominating things that come out of every discussion. Either somebody is talking about how they have automated infrastructure or they 1:09:55 automation in a way that it’s repeatable and it could be measured.


And then obviously there is the whole discussion about metrics and measurement, if you are not doing that, you are just wasting time. And then there is the whole cultural aspect. So yeah, it’s cool stuff.

Michael Coté: So speaking of measurement John, this is another one of my corny segues that I like to point out, because it has nothing to do with measurement. You were reminding me that Eucalyptus got $20 million in funding this week as well, I think like a B Round or something, right?

John M. Willis: Yeah, that would be a B Round.

Michael Coté: They have the Open Star famousness guy, or to use the term that is very popular among the DevOps people, rock star; that’s starting to grade on me, but whatever the kids like saying, I love it.

But anyways, Mårten Mickos is the CEO over there now for MySQL and everything. And in the Open Source world, he has been kicking up some dust by talking about like Open Core is the new model, and he is kind of like, this is a whole other — this is for the Open Source, like Torres kind of crowd angle here, is like to get really interested in the Open Source model and the business model and stuff.

There has been a whole lot of interesting stuff around Eucalyptus recently, in the sense that they have been — they have just been doing a lot. Or they have been involved in a lot of conversations I guess I should say that I have been in. There is lots of people who are going after an Open Source cloud thing, like there was just — it was yet another company that renamed itself while it launched, I forget what the new name is, it’s like Nymbler or Numbuler or something like that, and they used to be called some other unpronounceable name, and from some ex-Amazon guy.

John M. Willis: Well, it’s funny, because that’s — that guy, who founded that company, was Chris Brown; he was Chris Brown, who was the VP of Engineering in Opscode’s boss, and he is the guy — him and Chris Brown are the guys that went down to South Africa to start Amazon.

Michael Coté: And there is actually — you have probably seen the — like it’s one of the TechTarget pieces that talks about Opscode and the company who’s name I can’t remember how to say, that starts with an N.

But anyways, there is like three or four like Open Source cloud thing running around now, including Red Hat, and it’s funny to talk with all of them, they all like trash each other.

John M. Willis: Yeah, tough one.

Michael Coté: Like, what are you going to do, right? Of course they are your competitions so you say that they are — if you are very polite, as my seventh grade English teacher used to say, blowing out my candle won’t make yours brighter, so if you are very polite, you just point out how bright your candle is. But none of them are really very polite, they like to trash on each other, which is exciting.

John M. Willis: Well, it’s a bloodbath space man, and I tell you what, it’s — I mean, I saw that coming when I was over at the last company I was consulting at. I mean, that space is a bloodbath. I mean, it’s just — the thing is, is that hypervisor as a rule, hypervisor abstractions are not that hard.

I mean, I have seen a lot of people build fairly sophisticated clouds, if you will, private clouds on their own. You know what I mean? I don’t know. It just seems like, God bless Eucalyptus and you have got $20 million included and the guy running the company is lot smarter than I am, but it’s just — it’s a very — you have got Andy Bias doing like some really clever stuff with cloud scaling. And then you have got OpenNebula and you have got — there is just a lot going on there in the private cloud space, and that’s not even counting where’s dust eventually settles and then you have got — oh, by the way, IBM.

Michael Coté: That’s right. I am still waiting for IBM to like announce their like, here is our complete cloud stack thing. I mean, they definitely have — they have comprehensiveness and they have got their Dev Test thing out as a public cloud thing and everything, but they are not quite — I don’t know, they are very — being all encompassing in a technology sector can be as confusing as being no encompassing.

John M. Willis: You know what, I read the IBM Cloud for a long time, that Test Dev cloud was the real deal, and I suspect they are going to have some real deals coming up.

Michael Coté: No, I mean that’s the — we even talked about this when it came out, I mean that’s the obvious — that’s like the mature, stable way to do something in this area, is to release it as like, oh, this is for development and testing, and then you work out all the weird bugs and everything, and then you do it for production.

So I think you don’t have to be sort of like a person who splits open a bird and looks at their guts to figure that out, right? Whatever they call that, osprey or something.


So the last thing I was going to mention is, I have no details on this, but I have been talking with some — I think I mention this somewhere, but I have been looking at putting together like some little like panels here in Austin about just people talking about using cloud stuff. So I have been talking with a lot of people locally about that.

Someone mentioned to me that actually the Austin CloudCamp, which I missed, because it was in the evening, I will be frank, and I wanted to spend time with the family, that there was actually a really good turnout here in Austin, and there was a lot of — there were very technical people like talking about hypervisors and things like that, which is exciting.

So there was a group of people, I think, kind of headed by Pervasive of all people, if you remember that outfit, and they are looking to start an Austin cloud users group. I think sometime in July — I should look up the date, because they actually gave it to me, but there is going to be the first meeting in July and I am going to give like a 30 minute, like state of the cloud talk to open it up, and that will be exciting.

But it will be, hopefully — you have had like some 8Ws user group stuff over in Atlanta and everything, but hopefully we can kind of pull together the cloud community here. It would be great, because there is actually — and talking with people about like wanting to do some stuff, there is a whole lot of interesting cloud people around here, not to mention rackspace cloud folks being here and all sorts of stuff.

So maybe we will get you to come down here and show your kid on the strike thing, that would be great.

John M. Willis: Sure. That would be good. 76:18. So hey, I don’t know if we are wrapping up, but like we got accepted at DevOps, right?

Michael Coté: Oh yeah, I forgot about that.

John M. Willis: We are going to be down in Antwerp, right?

Michael Coté: Yeah, it’s in November, right?

John M. Willis: Yeah. So we will podcast from there, drinking Belgium beer.

Michael Coté: Yeah. We can call it the waffle cloud, because I think Belgians love it when you reduce their whole culture down to waffles, as I recall they are into that.

But I must say, and I will have to check, but the last time I was in Belgium, which was in the mid-90s, I do remember being struck by the fact that we got off of the plane and we were walking to the airport and they had a vending machine with waffles in it, and that might have been for tourists, but it did leave — it did seem like there is a lot of waffling going on there.

John M. Willis: It leaves an interesting taste in your mouth.

Michael Coté: That’s right. But yeah, it was actually kind of — I am embarrassed to say, I never remember anyone’s names, as you know John, I am sure longtime listeners know. So I forget the gentleman’s name, was it Gee or something like that or anyways, or Guy as we would say in the Midwest.

But anyway, so this guy, who is one of the organizers, reached out to us — it was you, right?

John M. Willis: Yeah.

Michael Coté: And he wanted to do some sort of cloud thing and John sort of expanded it to include me, because I told him I need to get some — I wanted to do some international stuff.

John M. Willis: Oh, actually, I pulled a fast one, they wanted you and I figured, oh shoot, they might not want me this year.

Michael Coté: So we are just going to do a little like, I don’t know, we will talk about — we actually did submit an abstract, so we have something we are supposed to talk about. But we are just going to talk about DevOps essentially and how it kind of fits with them.

And the crowd that’s there — I mean, it’s a Java conference essentially, and you have been to DevOps before. We are kind of going to bring cloud to a bunch of European developers essentially.

And then I have another session that’s — and basically it’s a reprise of the talk I gave at the Philly Emerging Technologies for the Enterprise thing about how cloud helps with Agile development.

But yeah, that will be exciting. We will be doing that stuff. Maybe we will arrange — this is like in November, so that’s like — it might as well be five years from now as far as I am concerned. But we will have to arrange some sort of like little live recording and see if we can get some people together. That would be exciting.

John M. Willis: Sounds good.

Michael Coté: Well, do you have anything else to go over, John?

John M. Willis: No, actually I think we have been on for a pretty long time today.

Michael Coté: Yeah, yeah. According to my matrix, my MAC, if you will, its been 1 hour and 17 minutes.

John M. Willis: Oh, that’s not horrible.

Michael Coté: No, it’s not too bad. Alright. Well, we will see everyone — actually, next time, if it works out, largely due to me, we have had a lot of scheduling like — I am just difficult.

John M. Willis: I mean, like missed a DMTF — talking about like this changing, how the world is going to look different, I mean we are going to have one of the guys that owns one of the — like really important silos of this whole new changing world.

Michael Coté: We are going to have Winston Bumpus on from the DMTF.

John M. Willis: We are going to have to get some ITIL guy and maybe an IT skeptic and we can start this whole DevOps. But part of it has got to be, what does DMTF mean?

Michael Coté: Yeah. Maybe for the next DevOps thing that I can weasel out of in the John-Coté style, we should have a cloud 79:53 chromogen panel, that would be exciting.

John M. Willis: There you go!

Michael Coté: But on that note, we will see everyone next time.

Disclosure: Eucalyptus and IBM are clients, as is OpsCode where John works. See the RedMonk client list for other relevant clients.

Categories: Cloud, IT Management Podcast, Systems Management.

Comment Feed

2 Responses

Continuing the Discussion

  1. […] that we’re following at the moment – things like the conflation of duties that is dev/ops, the emergence of open source as a mainstream development mechanism, and the accelerating value of […]

  2. […] trends that we’re following at the moment – things like the conflation of duties that is dev/ops, the emergence of open source as a mainstream development mechanism, and the accelerating value of […]