 Hello, hello. Can you guys hear me? Great. Are we ready? Okay. I'm Gretchen Curtis. I'm a co-founder and chief strategy officer of Piston Cloud Computing. And this is the analyst perspective on OpenStack from the outside looking in. So why don't you guys go ahead and introduce yourselves. My name is Krish Subramanian. I'm the principal analyst of a booty analyst firm called Rishi Dot Research. As a disclosure, I have a bias towards open source. Hi. I'm Gary Chen from IDC. I'm a research manager there and I cover server virtualization software and something we call cloud system software, which is something like OpenStack. Hi. I'm Sean Michael Kerner. I'm a senior editor at Internet News, which is the new service of Quinn Street, a bunch of websites, eWeek, Internet News, Datamation Server Watch, eSecurity Planner Enterprise Networking, and I manage a site called Linux Today. And my full disclosure is I have a strong bias toward Linux. And I'm Stephen O'Grady. I'm the co-founder of RedMonk, a developer focused analyst firm we've been doing. Cloud for really as long as it's been a term. All right. So I have a few questions prepared ahead of time, but what I'd like to do is start off with those and then open the panelists up to questions from the audience. But before we start, could I see a quick show of hands, how many people we have here that are press? Ha-ha. You were trying to hide. How about analysts? Any analysts? OpenE and the ecosystem. Any developers? Best of you guys. Operations. Cool. All right. Okay. So first question. What are people telling you guys about OpenStack? What would you characterize the buzz as being mostly positive, mostly negative? And how much of this feedback is coming from press and Twitter versus actual customer interviews? Both positive as well as negative feedback from customers as well as folks on Twitter. When I say folks on Twitter, it's not some random group of people. I follow a select group of operations people on Twitter who are, we are a close-knit group there and I get their feedback. I feel that I get either overwhelmingly positive feedback or total negative feedback like it's a piece of crap. I cannot install it. I can't get it done. So right now it's both positive as well as negative. That's what I hear. Yeah. I think for me, I definitely hear both. But I think it's mostly on the positive side that people are excited that there's a new technology out there. There's a new community project. And there's certainly some rough spots and things along the way and if it's early code, maybe we'll have problems with it. But I think people are fairly optimistic for the future that it will be a really interesting technology. But I think the main way I would characterize it among people who are not like... Most of you are really involved in the OpenStack community so I think you understand it really well. You understand OpenSource really well. But I mean, I talked to a lot of enterprises that don't really understand OpenSource. So I think it's a lot of the things that we went through with Linux. How do I start using it? Who's doing it? I tried it out and it didn't work but they were just downloading kind of raw open source from the logic page, that kind of thing. So I think there's a lot of people that are just not really... I mean they may use OpenSource but I don't know if they really understand that process and the difference between community and commercial versions and things like that. For me, on my side of the fence, I'm seeing... I'm high on OpenStack but it's at the crescendo of the hype cycle where 8 out of 10 press releases, I see where anyone mentions the word cloud which is a requirement to beat my spam filter has the word OpenStack compatible in there and OpenStack is a metaphor for anything being open source and most of these companies are not. So it's the... for me and the inbound stuff that I see OpenStack today is the information superhighway of 2001. It's everywhere, it's anything, it's everything though of course everyone in this room knows better. And I think from my perspective, this is a comparison I've made a couple times actually probably with some of the people in the room, you know, we see OpenStack as not being terribly dissimilar from MySQL in the sense that there's a lot of buzz about the size of the community, there's a lot of buzz about the size of the ecosystem and there is a... you know, we feel a fair amount of complaints about the technology. Many of those technology sort of complaints have been addressed from release to release but I think that the criticisms that we do see tend to be in a relatively technical nature, you know, they'll point to, you know, R-Sync being for core parts of technology, things like that, you know, but that is, I would say sort of largely drowned out by positive feedback on the size and sort of momentum and number of participants. Just to cut it on you for a second because with MySQL, I know the Oracle guys love MySQL, of course. When I talk to Oracle, though, they're not high on OpenStack, though, but they are high on MySQL. I don't know how that relates, but... Yeah, I mean, the point wasn't to actually try to make a direct line comparison between the two projects but rather the kinds of feedback that they tend to gather publicly. So a common criticism of OpenStack is that it's a commercially-driven, open-source project as opposed to a developer-driven project and that the marketing around the project has far surpassed any development work and has been far too aggressive for where the project is from a technical standpoint. So in your opinion, is this true and what would be the best way for the community to address this criticism? I'll just jump in. Some of that criticism comes from the Linux side where people are used to Debian and Slackware so everything started in the community and then companies came first. Here it's the other way around. So that's just the historical context. When I see it on the press side, because we all grew up with Debian and Slackware and then Red Hat afterwards and then Soussi and then Ubuntu, et cetera, here it's going the other way. In my case, like I also, as he said, I come from Linux era, so I thought there is way too much marketing going on. So I did a post saying probably developers should get AppleHand and not. Immediately after I published that post, I got pushed back from lots of people who are not associated with RackSpace or OpenStack. They told me that you need that kind of marketing, especially when the competing players like VMware and Cloud Citrix are unleashing that kind of bullshit against OpenStack. We need this kind of marketing to sort of neutralize that anti-OpenStack campaign that's going on. So probably I see their point and in my opinion there should be a proper mix of marketing as well as developer voice. I'm worried that sometimes developers feel frustrated that marketing takes more precedence over their own voice. I guess keeping a level of transparency might help. Well, I'm sure that we all, we probably all would address that differently from our perspective. RedMonk is sort of most focused on developers, most focused on practitioners. So we obviously talk to many of the vendors involved and they'll brief us on products and so on. But at the end of the day, the most important feedback that we get is either directly from developers or we will mine quantitatively a variety of developer-related sources. So this will give us hopefully sort of in an ideal situation with qualitative and quantitative feedback in terms of the people actually using the technology as opposed to the marketing. I think on a larger question, really the only difficulty that I would see with too much marketing is that people show up and their expectations aren't met and then they even don't come back. That's really a part from that. It's very difficult to have too much marketing because that's like having too much visibility. How is that a problem? And thus far, at least with OpenStack, we really haven't seen that. It's not as if you had an initial way of adopters. Everybody showed up and left. Over the past couple of years, you've had more and more people showing up. Yeah, I agree with Stephen. I don't think you can have too much marketing. I mean, the cloud battles are happening. Everyone's fighting for an ecosystem, fighting for developers. There's always going to be some hype buildup that you're not going to be able to control and people like Stephen said are going to have expectations that might not be met at some early stage. But I think that's just something that just kind of comes along with having a highly visible project. I think the whole individual developer thing, it is going to be a lot different. Stuff like Linux, even some of the Android enthusiast community or something like that. That was something that individual people worked on as hobbies and something that they kind of used themselves personally. I don't know if it will be a similar thing because I don't know how many people are going to want to run a cloud at home or something like that. It might entice a different sort of developer community. Something that's more commercially focused. I don't know if that's necessarily bad. It's just that that's what the market is for and that's who the participants are. I think if there's a diverse amount of participants then it could work out. I think the good thing about Linux, if you look at Linus, how he runs Linus, he's not really tied to any commercial organization. He's kind of like the benevolent dictator. He can do whatever he wants. That's done interesting things with a project. I think it's going to be a little different, but I think it could still be a good model. I think taking from that, I would say, since we have moved from scratching your own edge to doing some, doing a lot of commerce activity around open-source software, I think having some decent level of marketing is necessary. But at the same time, probably it would be better off if we have Linus Torvalds like a benevolent dictator in the open-stack community who takes communities cause and matches with the business requirement. Probably that will sort of solve some of the developers' concerns. It's vendor-driven, mostly not developer-driven. At least the criticism was like it's getting to a point where it's more of vendor-driven. My question is like, is it necessarily bad or good? What is your point on that? Just to interrupt there, I would say it should be user-driven to Jonathan's point this morning because users feed the developers, the vendors who look for the money and it's just a virtuous cycle. And just to go back because I have to get my two cents and hear the sound of my own voice, identifying what the bullshit is from what's real on my side of the fence, because I don't talk to as many users. I'm a little bit abstracted from that. It's relatively straightforward. Some people don't run clouds in their house, but thanks to the magic of Linux, I can. So that I can test and I try out everything. Open-source is the great validator because it removes any barriers to adoption and makes it very easy to try. And then for commercial, there's only one way anybody can ever do it, and that's lighthouse wins. You have, or gate house customers, or lighthouse customers, however you want to call it. So that way you see, okay, this is used in production, this is real, right? OpenStack is real, people are using it. It's not just some pie in the sky, hypothetical Microsoft surface product that may or may not be available in 30 days, right? We'll see. So to your question in terms of sort of vendor versus developer-driven, I actually think to some degree that's a, it's a, it's almost a mood point, right? So in other words, I don't know how many of you, you know, saw that the quote at the time, Paul Moritz, you know, who was at the time CEO of VMware, you know, was at an event in London, and you know, this is the CEO of VMware talking, and basically said, look, if we don't give developers what we want, we're going to be irrelevant, right? So at the end of the day, I don't care if it's coming from vendors, I don't care if it's coming from users, I don't care if it's coming actually from the developers themselves. If developers don't want to use it, they won't. That's more or less the world that we live in today, and it's becoming more so all the time. So in other words, you know, if we have initiatives, you know, that are more vendor driven, and certainly we've seen these before, right? The Eclipse Foundation has been largely driven by, you know, consortium of vendors and has produced more or less consistently developer-friendly tools over time, you know, there are certainly exceptions. You know, I mean, I don't want to sort of dismiss it as a concern, but I would say over the longer term, you know, any consortium foundation or otherwise it isn't producing tools from developers is not going to be relevant. And as I said, I think based on the adoption that we've seen from OpenStack today and the interests, that would not be my primary concern. I'll put it that way. Yeah, I mean, I kind of agree with Steven. I mean, I don't know vendor developer, but I think the cases that you have to focus on, you know, okay, what happens if there's conflict, right, and someone has to make a decision, right? And in Linux, there was some guy, you know, if it just finally got to it, he said, okay, this is going to be the way it is, and that's the way it was. And then, you know, who is going to set kind of, you know, this strategic direction or vision, right? And I think some things like that aren't necessarily done by a committee so, you know, maybe the foundation will have some role in that. Yeah, I agree with all of them that I don't care if it's vendor-driven or developer-driven. But regarding the criticism against OpenStack being a vendor-driven community, I want to look at it from a totally different angle. Like Linux, it was mostly user-driven because they were catching their age and users were using the system. And that's why we saw a lot of users contributing to it. But now we are moving from taking care of the users to taking care of the needs of the enterprises and service providers. It's a completely different dynamic here. So, when it happens, there will be participation from the vendors, there will be participation from the service providers and enterprises. So, like, you will see more employees of these organizations coming into participation than end users or developing the software as a hobby. So, I think there's a change in the dynamics and people need to understand and appreciate the fact before they sort of criticize when they do an operation. Just to talk about Linux for a second, it's important to realize that Linus doesn't actually code anymore, right? He doesn't actually introduce new technology. He's the gatekeeper, the traffic manager and the guy who's really good for guys like me because he calls, you know, storage Satan and evil companies like crack organization, all kinds of other things. It is the vendors in the Linux community and always has been the vendors that drive that forward but there is that, you know, benevolent dictator at the top, which is really quite amazing. Now, Eclipse is really interesting because Eclipse, and you know, Mike Malinkovich as well as I do, exists as an organization to build open source software on which vendors like Intel SAP etc. can then commercialize software. If that is the extra goal of the OpenStack Foundation to have, you know, an OpenStack base on which vendors can build commercial software then so be it. I don't know that it is or isn't. I think OpenStack can and will stand on its own. There are very, very few people that I know that use Eclipse core and it's actually there are some for Android but usually they're using something on top, an extension commercial extension always. That's just how you use any Eclipse project. Probably I'll take this opportunity to vent my frustration about the pundits out there. The thing is they all hate Richard Stallman because he's a communist. But when they where their pundit had, punditry had, they talk like Richard Stallman saying that hey, vendors are bad and all those things. I think there's some level of either lack of understanding or hypocrisy involved here. So I think pundits who are criticizing OpenStack kind of projects for vendor involvement should rethink their strategy. Openness is to open stack like conservation is to conservatives. Krish. Do you want to talk a little bit about it? Let me first make it clear that it was a tweet I just made after hearing that from somebody because I like that tone of that tweet so that's why I tweeted it. It was not my personal opinion but having said that I think there is a level of marketing going on about openness in OpenStack. I keep hearing from people that if you install OpenStack that means you have a perfectly open system available to you and you're not logged in and all those marketing things. But let's take the case of HP so they have taken OpenStack but they have built some proprietary they have added proprietary code to it which they haven't contributed back to the community. So when you use their services it is not as easy as taking it from a provider who is using vanilla OpenStack distro and moving it to another vanilla OpenStack distro. There is some level of work involved, some level of cost involved and these things are getting washed under the marketing that's going on here. So that is something which you need to take into account but having said that it's much better than having a proprietary software and being completely clueless and getting logged in there. I just wanted to warn people that there is quite a bit of marketing going on about openness in the OpenStack ecosystem. I think this whole OpenCloud thing who knows what an OpenCloud is I mean how do you define it and I think in a lot of ways I don't know if the OpenCloud really exists in the terms that people say well what does that mean everything is an official standard and it's OpenStack should be compatible with this cloud and that cloud and that cloud and I don't know if those things are really realistic. Have we ever had any of that kind of compatibility in IT? Compatibility has already been really difficult even between different flavors of Linux and it's still better going from Linux to Linux and Linux to Windows but no one ever said hey Linux you should really be a lot more portable and make yourself interoperable with Windows and there's only so much you can do and who's going to drive that forward the standards organizations aren't really going to do it or they're not going to do it fast enough so I think when you look at projects like OpenStack if you think of openness that's probably about as open as you're going to get let's make this stuff make it open source get all these people to contribute whoever wants to contribute and we'll try and make it work with as many things as we can or who wants to let us make it work with it so I mean that's probably as best as you're going to get so I think it really depends on what you're talking about when you mean open but I think the word is so abused now doesn't really mean anything and just to pick up on that I would certainly agree with the statement that compatibility is a relatively loose term so in other words anybody in the room ever do J2E implementations? A handful of people in the room those people probably know that if you take two J2E platforms and try to move an application from one to another a whole bunch of things are going to break and that's sort of with the NA standard so that's sort of understood I think most people in the industry will understand that you go from one version of Windows, for example to another, some things break same with Linux, same with Mac and so on all of that said I think that with OpenStack is basically licensing so Linux obviously has a reciprocal style license which means that I can't ship and distribute a sort of materially different sort of well I shouldn't say that I can ship something different but I have to release my changes under exactly the same license so anything I improve anything I change, anything I edit has to be made available under those same terms so that basically everybody has the opportunity to work off the same platform I'm sure most of you in the room know is permissively licensed and so I have no such obligations from a licensing perspective which allows me to do different things so the question I think is going to be for the community, for the vendors, for the users and ultimately for the developers is that going to be a problem and if it is a problem how do you go about solving that we've solved problems like this before J2E as much as I just bashed it was a workable solution to that problem and ultimately port with a minimal level of frustration from one platform to another so the last thing I ever want to do is recreate J2E but if we end up having this problem are we going to see for example standards emerge which are if you're adhering to this subset or this superset of APIs great you carry the open stack certification if you're not you don't have the certification so on I just wanted to put in one thing before we jump in on that open in a world where a company like Oracle which locks everybody in the database and Microsoft which locks everybody in and everything can claim to be open it means nothing right but from my experience open means only one thing in the cloud world and that's workload portability when VMware which may or may not be an open company I can take my workload VMotion but that's what people care really it's that workload portability if I can move it back and forth private, public, hybrid here, there, back, forth, sideways that's the matter licensing is a marvelous topic I believe that GPL is mother, father and everything in between I know open stack will never go GPL and that's fine and nice we should get Richard Stallman here one year but workload portability is what matters to operations people I think I'll take a crack at the first one I think ultimately you know are sort of the name and trademark are they sufficient no I don't believe they are in the sense that there are and this is not through the fault of the folks in the room who are working on it or the folks in the room who are employed by the foundations it's simply a fact that everybody in here who's implemented technology knows that there are literally tens of thousands of moving pieces and inevitably when you have deltas between implementation and another there are things that are going to break there are things that are not going to work exactly as advertised and so on that is inevitable I don't care how hard you try and I don't care I mentioned J2E they spent a lot of time on that specification trying to get it perfectly compatible and it was not so ultimately I guess what I'm saying is yes is that a step in the right direction in the sense that if you have a clearly incompatible version that is clearly breaking things across the board of protecting your brand by saying trademark and so on is that going to be sufficient to guarantee the kinds of workload portability or compatibility, cross-platform compatibility that people are looking for probably not you're going to need something that's more sophisticated and more unfortunately complicated in the longer term I think you have to look at things like compatibility as kind of it's not just a yes or no it's kind of a shade of gray it's like how much pain are you going to go through right and so you want to make it a reasonable it may not be something that you would love to do but it was something that it's feasible if you really had to do it and it wouldn't be too painful so I think if you're looking at maintaining the open stack brand I think there should be some kind of minimum level of compatibility in order to use it you have to implement these core set of APIs maybe there's even some kind of qualification we're going to test it make sure it really works and really behaves I know Android does something like that to be called Android you have to run through these compatibility test suites just to make sure that if I do this operation that it really behaves in a similar manner I don't think you're going to you don't want to go too far on that too because you don't want to stifle innovation and people can't do anything but I think you have to define a core level and make people say you know at least you can adhere to these core things and maybe implement a testing program to actually validate that there is compatibility in terms of lowest common set of services so I think sort of final question before we wrap up here what is the biggest challenge that you feel open stack will face next year and do you see any major holes in the ecosystem the challenge is coming from two fronts like on the perception side there is a perception that the corporations involved in the open stack project is controlling the project and it's going to sort of inhibit the true spirit of openness and all those things so I think open stack really needs to do something to change that perception it's partly because of the anti open stack campaign that has been going on in the industry and some of the things I am seeing from the open stack side for example in the last board meeting there was some talk about keeping the discussion secret so I think transparency is the key if you really want to take out the perception that's in the industry that there is a certain level of corporate control that is going on so it is critical that open stack makes transparency the core thing in everything they do and that is the only way they can gain community credibility and get rid of the corporate control backlash the other one is on the technological side one of the biggest challenges which open stack faces at least that's what I hear from our clients as well as from people I talk with I think it's about ease of deployment that are working on deploying open stack seamless but I think that is something which you need to focus on before you see large scale adoption I think a couple of challenges I think on the community project front is deciding where to take the project there is a lot of people that want to do a lot of different things with it I don't think you want to go too far out of scope on certain things I think it has to be a decision made in terms of cloud I have seen cloud turning into a lot of different things there is Amazon style cloud where we build everything from this blank sheet of paper for a new set of applications and then there is clouds that are meant to do a lot of stuff with all the existing workloads that are out there and I have this piece of hardware installed and I want to use it do you want to get into existing workloads that are out there and hardware which can be a huge thing or do you want to try for green field new style applications and not be way down with legacy I think that is one thing to look at and I agree with Chris I think the main thing I would look for next year is really the commercialization of open stack and with commercialization I think it really brings a wider audience today it is really I think cloud providers people whose IT is their business it is core to their business so they spend a lot of time in technology they have expertise to manage it and use it but there is another set of enterprise market where IT is not their business it enables their business so they are not really looking to build their own open stack and they are looking for something that is really consumable how do I make it really easy to deploy and update and manage and who is going to do security and back port fixes and decide which release are solid releases that should go production and things like that so I think that will alleviate a lot of the frustration that people have had where today if you wanted to do it it is a largely do it yourself kind of thing I don't think that is for everyone so I would like to see a lot more commercial activity next year I see only two challenges fragmentation and long term support these are the same two challenges that Linux faced at the same stage and Android is facing the same challenges too you think the user base on Android there is people running 2, 2, 3, 4 and everything in between as a developer I know I have a target of six different versions different screen sizes it is all different if open stack has that same kind of fragmentation it runs down the same path 15 years ago where once upon a time I could just have Unix system 5 and then you know we have AIX Solaris and everything in between and they are all different but not completely related and it is challenging the long term support issue the Red Hat guys and I don't think there are any Red Hat guys in the audience but there might be they talk to me about this all the time because they like 5, 10 years support that is what they like to offer to their customers so if they are going to base their Red Hat Enterprise open stack release and say Folsom and let's say they are going to support that for 3 years, 5 years and then they are going to back port all these fixes back what does that mean for fragmentation and long term support for everybody else that is going to be an interesting challenge as there will be different versions, different things and I think this innovation that came in in this last cycle with Quantum and I hear a lot of talks where people are moving from Nova compute and the Nova networking stack to Quantum that kind of shift in a long term support model to be possible from the early adopters that help to drive this great acceleration in the last 2 years so that will be a real challenge as you get more adoption I would basically just agree with that I think there are any number of challenges for the project that we could talk about certainly I know the Gardner piece was referenced this morning and some of the critiques that they had and so on particularly about the technical aspects of the project to me those are all solvable problems same with Christian's perceptions of vendor involvement and so on that is a concern but again I think that is a solvable problem the long term challenge for me is fragmentation I think it's going to be very interesting to see how the project evolves moving forward because the sort of volume in number of participants at some point could be a double-edged sword because right now it's great it's momentum, you have contributions flowing in, you have more interest you have more visibility, more marketing and so on but the fact is that at the end of the day a lot of these parties are going to have very different agendas and in many cases competing agendas so are we going to end up to borrow Sean's android analogy with one android flavor and then any of you who have android flavors that come from different vendors with the crap that they install on it completely different experiences based off the same pieces which experientially that may be fine but if you then have issues with porting workloads that's a problem so anything that the folks in the audience can do to sort of target and stave off fragmentation I think that to me is the biggest challenge Clark is taking how? I agree with that I don't know exactly who said it, how to year time is moving by the danger in situations like that typically becomes when you compete with a ubiquitous incumbent you're screwed or you're not screwed it just takes much longer I don't see that here I don't see really anybody running away with a market to the standpoint that OpenStack if it's not exactly where it needs to be 6 months or even 12 months from now is doomed but I mean it's really a fast moving industry so in that sense you need to move pretty quickly I think you're talking about opening generation in the OpenStack community the article that came in the network would probably a few years back I think it's a short sighted comment from the analysts and the analysts I responded to the article there itself but the analysts didn't differentiate between the billions of dollars spent by Amazon, Microsoft and VMware compared to probably a few millions or even less spent by vendors in the OpenStack community and there is no real pressure to make money within a few months so I think he has got it wrong